chore(wasm): update dependencies of wasm tests

Using the functions that were introduced recently,
it is possible to make the full_propagate_parallelized method
more parallel than it was, resulting in faster computations.

The new carry propapagation should now be the cost of
a default add + one PBS, so ~400ms in 256 instead of ~3s.

However its probably slower for smaller number of blocks (eg 4 blocks)

This is done by extracting carry and messages in parallel,
then adding the carries to the correct message, the final step
is to use the single carry propapagation function.

.config/nextest.toml

@@ -8,10 +8,10 @@ slow-timeout = "5m"
 
 
 [[profile.ci.overrides]]
-filter = 'test(/^.*param_message_1_carry_[567]$/) or test(/^.*param_message_4_carry_4$/)'
+filter = 'test(/^.*param_message_1_carry_[567]_ks_pbs$/) or test(/^.*param_message_4_carry_4_ks_pbs$/)'
 retries = 3
 
 [[profile.ci.overrides]]
-filter = 'test(/^.*param_message_[23]_carry_[23]$/)'
+filter = 'test(/^.*param_message_[23]_carry_[23]_ks_pbs$/)'
 retries = 1
 

.github/pull_request_template.md

@@ -0,0 +1,13 @@
+<!-- Feel free to delete the template if the PR (bumping a version e.g.) does not fit the template -->
+closes: _please link all relevant issues_
+
+### PR content/description
+
+### Check-list:
+
+* [ ] Tests for the changes have been added (for bug fixes / features)
+* [ ] Docs have been added / updated (for bug fixes / features)
+* [ ] Relevant issues are marked as resolved/closed, related issues are linked in the description
+* [ ] Check for breaking changes (including serialization changes) and add them to commit message following the conventional commit [specification][conventional-breaking]
+
+[conventional-breaking]: https://www.conventionalcommits.org/en/v1.0.0/#commit-message-with-description-and-breaking-change-footer

.github/workflows/aws_tfhe_fast_tests.yml

@@ -0,0 +1,119 @@
+# Run a small subset of shortint and integer tests to ensure quick feedback.
+name: Fast AWS Tests on CPU
+
+env:
+  CARGO_TERM_COLOR: always
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+  RUSTFLAGS: "-C target-cpu=native"
+
+on:
+  # Allows you to run this workflow manually from the Actions tab as an alternative.
+  workflow_dispatch:
+    # All the inputs are provided by Slab
+    inputs:
+      instance_id:
+        description: "AWS instance ID"
+        type: string
+      instance_image_id:
+        description: "AWS instance AMI ID"
+        type: string
+      instance_type:
+        description: "AWS instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: 'Slab request ID'
+        type: string
+      fork_repo:
+        description: 'Name of forked repo as user/repo'
+        type: string
+      fork_git_sha:
+        description: 'Git SHA to checkout from fork'
+        type: string
+
+jobs:
+  fast-tests:
+    concurrency:
+      group: ${{ github.workflow }}_${{ github.ref }}_${{ inputs.instance_image_id }}_${{ inputs.instance_type }}
+      cancel-in-progress: true
+    runs-on: ${{ inputs.runner_name }}
+    steps:
+      # Step used for log purpose.
+      - name: Instance configuration used
+        run: |
+          echo "ID: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+          echo "Fork repo: ${{ inputs.fork_repo }}"
+          echo "Fork git sha: ${{ inputs.fork_git_sha }}"
+
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: ${{ inputs.fork_repo }}
+          ref: ${{ inputs.fork_git_sha }}
+
+      - name: Set up home
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install latest stable
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: stable
+          default: true
+
+      - name: Run core tests
+        run: |
+          AVX512_SUPPORT=ON make test_core_crypto
+
+      - name: Run boolean tests
+        run: |
+          make test_boolean
+
+      - name: Run user docs tests
+        run: |
+          make test_user_doc
+
+      - name: Run js on wasm API tests
+        run: |
+          make test_nodejs_wasm_api_in_docker
+
+      - name: Gen Keys if required
+        run: |
+          make gen_key_cache
+
+      - name: Run shortint tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE FAST_TESTS=TRUE make test_shortint_ci
+
+      - name: Run integer tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE FAST_TESTS=TRUE make test_integer_ci
+
+      - name: Run shortint multi-bit tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE FAST_TESTS=TRUE make test_shortint_multi_bit_ci
+
+      - name: Run integer multi-bit tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE FAST_TESTS=TRUE make test_integer_multi_bit_ci
+
+      - name: Run high-level API tests
+        run: |
+          make test_high_level_api
+
+      - name: Slack Notification
+        if: ${{ always() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Fast AWS tests finished with status: ${{ job.status }}. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/aws_tfhe_integer_tests.yml

@@ -0,0 +1,86 @@
+name: AWS Integer Tests on CPU
+
+env:
+  CARGO_TERM_COLOR: always
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+  RUSTFLAGS: "-C target-cpu=native"
+
+on:
+  # Allows you to run this workflow manually from the Actions tab as an alternative.
+  workflow_dispatch:
+    # All the inputs are provided by Slab
+    inputs:
+      instance_id:
+        description: "AWS instance ID"
+        type: string
+      instance_image_id:
+        description: "AWS instance AMI ID"
+        type: string
+      instance_type:
+        description: "AWS instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: 'Slab request ID'
+        type: string
+      fork_repo:
+        description: 'Name of forked repo as user/repo'
+        type: string
+      fork_git_sha:
+        description: 'Git SHA to checkout from fork'
+        type: string
+
+jobs:
+  integer-tests:
+    concurrency:
+      group: ${{ github.workflow }}_${{ github.ref }}_${{ inputs.instance_image_id }}_${{ inputs.instance_type }}
+      cancel-in-progress: true
+    runs-on: ${{ inputs.runner_name }}
+    steps:
+      # Step used for log purpose.
+      - name: Instance configuration used
+        run: |
+          echo "ID: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+          echo "Fork repo: ${{ inputs.fork_repo }}"
+          echo "Fork git sha: ${{ inputs.fork_git_sha }}"
+
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: ${{ inputs.fork_repo }}
+          ref: ${{ inputs.fork_git_sha }}
+
+      - name: Set up home
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install latest stable
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: stable
+          default: true
+
+      - name: Gen Keys if required
+        run: |
+          make gen_key_cache
+
+      - name: Run integer tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE make test_integer_ci
+
+      - name: Slack Notification
+        if: ${{ always() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Integer tests finished with status: ${{ job.status }}. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/aws_tfhe_multi_bit_tests.yml

@@ -0,0 +1,90 @@
+name: AWS Multi Bit Tests on CPU
+
+env:
+  CARGO_TERM_COLOR: always
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+  RUSTFLAGS: "-C target-cpu=native"
+
+on:
+  # Allows you to run this workflow manually from the Actions tab as an alternative.
+  workflow_dispatch:
+    # All the inputs are provided by Slab
+    inputs:
+      instance_id:
+        description: "AWS instance ID"
+        type: string
+      instance_image_id:
+        description: "AWS instance AMI ID"
+        type: string
+      instance_type:
+        description: "AWS instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: 'Slab request ID'
+        type: string
+      fork_repo:
+        description: 'Name of forked repo as user/repo'
+        type: string
+      fork_git_sha:
+        description: 'Git SHA to checkout from fork'
+        type: string
+
+jobs:
+  multi-bit-tests:
+    concurrency:
+      group: ${{ github.workflow }}_${{ github.ref }}_${{ inputs.instance_image_id }}_${{ inputs.instance_type }}
+      cancel-in-progress: true
+    runs-on: ${{ inputs.runner_name }}
+    steps:
+      # Step used for log purpose.
+      - name: Instance configuration used
+        run: |
+          echo "ID: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+          echo "Fork repo: ${{ inputs.fork_repo }}"
+          echo "Fork git sha: ${{ inputs.fork_git_sha }}"
+
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: ${{ inputs.fork_repo }}
+          ref: ${{ inputs.fork_git_sha }}
+
+      - name: Set up home
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install latest stable
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: stable
+          default: true
+
+      - name: Gen Keys if required
+        run: |
+          make GEN_KEY_CACHE_MULTI_BIT_ONLY=TRUE gen_key_cache
+
+      - name: Run shortint multi-bit tests
+        run: |
+          make test_shortint_multi_bit_ci
+
+      - name: Run integer multi-bit tests
+        run: |
+          make test_integer_multi_bit_ci
+
+      - name: Slack Notification
+        if: ${{ always() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Shortint tests finished with status: ${{ job.status }}. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/aws_tfhe_tests.yml

@@ -25,26 +25,35 @@ on:
       request_id:
         description: 'Slab request ID'
         type: string
-      matrix_item:
-        description: 'Build matrix item'
+      fork_repo:
+        description: 'Name of forked repo as user/repo'
+        type: string
+      fork_git_sha:
+        description: 'Git SHA to checkout from fork'
         type: string
 
 jobs:
   shortint-tests:
     concurrency:
-      group: ${{ github.ref }}_${{ github.event.inputs.instance_image_id }}_${{ github.event.inputs.instance_type }}
+      group: ${{ github.workflow }}_${{ github.ref }}_${{ inputs.instance_image_id }}_${{ inputs.instance_type }}
       cancel-in-progress: true
-    runs-on: ${{ github.event.inputs.runner_name }}
+    runs-on: ${{ inputs.runner_name }}
     steps:
       # Step used for log purpose.
       - name: Instance configuration used
         run: |
-          echo "ID: ${{ github.event.inputs.instance_id }}"
-          echo "AMI: ${{ github.event.inputs.instance_image_id }}"
-          echo "Type: ${{ github.event.inputs.instance_type }}"
-          echo "Request ID: ${{ github.event.inputs.request_id }}"
+          echo "ID: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+          echo "Fork repo: ${{ inputs.fork_repo }}"
+          echo "Fork git sha: ${{ inputs.fork_git_sha }}"
 
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: ${{ inputs.fork_repo }}
+          ref: ${{ inputs.fork_git_sha }}
 
       - name: Set up home
         run: |
@@ -58,7 +67,7 @@ jobs:
 
       - name: Run core tests
         run: |
-          make test_core_crypto
+          AVX512_SUPPORT=ON make test_core_crypto
 
       - name: Run boolean tests
         run: |
@@ -72,17 +81,21 @@ jobs:
         run: |
           make test_user_doc
 
-      - name: Run js on wasm API tests
-        run: |
-          make test_nodejs_wasm_api_in_docker
-
       - name: Gen Keys if required
         run: |
           make gen_key_cache
 
       - name: Run shortint tests
         run: |
-          make test_shortint_ci
+          BIG_TESTS_INSTANCE=TRUE make test_shortint_ci
+
+      - name: Run high-level API tests
+        run: |
+          BIG_TESTS_INSTANCE=TRUE make test_high_level_api
+
+      - name: Run example tests
+        run: |
+          make test_examples
 
       - name: Slack Notification
         if: ${{ always() }}

.github/workflows/aws_tfhe_wasm_tests.yml

@@ -0,0 +1,87 @@
+name: AWS WASM Tests on CPU
+
+env:
+  CARGO_TERM_COLOR: always
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+  RUSTFLAGS: "-C target-cpu=native"
+
+on:
+  # Allows you to run this workflow manually from the Actions tab as an alternative.
+  workflow_dispatch:
+    # All the inputs are provided by Slab
+    inputs:
+      instance_id:
+        description: "AWS instance ID"
+        type: string
+      instance_image_id:
+        description: "AWS instance AMI ID"
+        type: string
+      instance_type:
+        description: "AWS instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: 'Slab request ID'
+        type: string
+      fork_repo:
+        description: 'Name of forked repo as user/repo'
+        type: string
+      fork_git_sha:
+        description: 'Git SHA to checkout from fork'
+        type: string
+
+jobs:
+  wasm-tests:
+    concurrency:
+      group: ${{ github.workflow }}_${{ github.ref }}_${{ inputs.instance_image_id }}_${{ inputs.instance_type }}
+      cancel-in-progress: true
+    runs-on: ${{ inputs.runner_name }}
+    steps:
+      # Step used for log purpose.
+      - name: Instance configuration used
+        run: |
+          echo "ID: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+          echo "Fork repo: ${{ inputs.fork_repo }}"
+          echo "Fork git sha: ${{ inputs.fork_git_sha }}"
+
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: ${{ inputs.fork_repo }}
+          ref: ${{ inputs.fork_git_sha }}
+
+      - name: Set up home
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install latest stable
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: stable
+          default: true
+
+      - name: Run js on wasm API tests
+        run: |
+          make test_nodejs_wasm_api_in_docker
+
+      - name: Run parallel wasm tests
+        run: |
+          make install_node
+          make ci_test_web_js_api_parallel
+
+      - name: Slack Notification
+        if: ${{ always() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "WASM tests finished with status: ${{ job.status }}. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/boolean_benchmark.yml

@@ -5,28 +5,25 @@ on:
   workflow_dispatch:
     inputs:
       instance_id:
-        description: 'Instance ID'
+        description: "Instance ID"
         type: string
       instance_image_id:
-        description: 'Instance AMI ID'
+        description: "Instance AMI ID"
         type: string
       instance_type:
-        description: 'Instance product type'
+        description: "Instance product type"
         type: string
       runner_name:
-        description: 'Action runner name'
+        description: "Action runner name"
         type: string
       request_id:
-        description: 'Slab request ID'
+        description: "Slab request ID"
         type: string
-      matrix_item:
-        description: 'Build matrix item'
-        type: string
-
 
 env:
   CARGO_TERM_COLOR: always
   RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
 
 jobs:
   run-boolean-benchmarks:
@@ -40,14 +37,13 @@ jobs:
           echo "AMI: ${{ inputs.instance_image_id }}"
           echo "Type: ${{ inputs.instance_type }}"
           echo "Request ID: ${{ inputs.request_id }}"
-          echo "Matrix item: ${{ inputs.matrix_item }}"
 
       - name: Get benchmark date
         run: |
           echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
 
       - name: Checkout tfhe-rs repo with tags
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           fetch-depth: 0
 
@@ -62,35 +58,24 @@ jobs:
           toolchain: nightly
           override: true
 
-      - name: Run benchmarks
+      - name: Run benchmarks with AVX512
         run: |
-          make bench_boolean
+          make AVX512_SUPPORT=ON bench_boolean
 
       - name: Parse results
         run: |
           COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
           COMMIT_HASH="$(git describe --tags --dirty)"
           python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
-          --database tfhe_rs_benchmarks \
+          --database tfhe_rs \
           --hardware ${{ inputs.instance_type }} \
           --project-version "${COMMIT_HASH}" \
           --branch ${{ github.ref_name }} \
           --commit-date "${COMMIT_DATE}" \
-          --bench-date "${{ env.BENCH_DATE }}"
-
-      - name: Remove previous raw results
-        run: |
-          rm -rf target/criterion
-
-      - name: Run benchmarks with AVX512
-        run: |
-          make AVX512_SUPPORT=ON bench_boolean
-
-      - name: Parse AVX512 results
-        run: |
-          python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
+          --bench-date "${{ env.BENCH_DATE }}" \
+          --walk-subdirs \
           --name-suffix avx512 \
-          --append-results
+          --throughput
 
       - name: Measure key sizes
         run: |
@@ -101,7 +86,7 @@ jobs:
           python3 ./ci/benchmark_parser.py tfhe/boolean_key_sizes.csv ${{ env.RESULTS_FILENAME }} \
           --key-sizes \
           --append-results
-  
+
       - name: Upload parsed results artifact
         uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
         with:
@@ -109,7 +94,7 @@ jobs:
           path: ${{ env.RESULTS_FILENAME }}
 
       - name: Checkout Slab repo
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           repository: zama-ai/slab
           path: slab
@@ -118,13 +103,25 @@ jobs:
       - name: Send data to Slab
         shell: bash
         run: |
-          echo "Computing HMac on downloaded artifact"
+          echo "Computing HMac on results file"
           SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
           echo "Sending results to Slab..."
           curl -v -k \
           -H "Content-Type: application/json" \
           -H "X-Slab-Repository: ${{ github.repository }}" \
-          -H "X-Slab-Command: store_data" \
+          -H "X-Slab-Command: store_data_v2" \
           -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
           -d @${{ env.RESULTS_FILENAME }} \
           ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Boolean benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/cargo_build.yml

@@ -17,21 +17,21 @@ jobs:
 
     strategy:
       matrix:
-        os: [ubuntu-latest, macos-latest]
+        os: [ubuntu-latest, macos-latest, windows-latest]
       fail-fast: false
 
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
-
-      - name: Get rust toolchain to use for checks and lints
-        id: toolchain
-        run: |
-          echo "rs-toolchain=$(make rs_toolchain)" >> "${GITHUB_OUTPUT}"
+      - uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
 
       - name: Run pcc checks
         run: |
           make pcc
 
+      - name: Build Release core
+        run: |
+          make build_core AVX512_SUPPORT=ON
+          make build_core_experimental AVX512_SUPPORT=ON
+
       - name: Build Release boolean
         run: |
           make build_boolean
@@ -40,9 +40,13 @@ jobs:
         run: |
           make build_shortint
 
-      - name: Build Release shortint and boolean
+      - name: Build Release integer
         run: |
-          make build_boolean_and_shortint
+          make build_integer
+
+      - name: Build Release tfhe full
+        run: |
+          make build_tfhe_full
 
       - name: Build Release c_api
         run: |

.github/workflows/check_commit.yml

@@ -10,7 +10,7 @@ jobs:
       - name: Check first line
         uses: gsactions/commit-message-checker@16fa2d5de096ae0d35626443bcd24f1e756cafee
         with:
-          pattern: '^((feat|fix|chore|refactor|style|test|docs|doc)\(\w+\)\:) .+$'
+          pattern: '^((feat|fix|chore|refactor|style|test|docs|doc)(\(\w+\))?\:) .+$'
           flags: "gs"
           error: 'Your first line has to contain a commit type and scope like "feat(my_feature): msg".'
           excludeDescription: "true" # optional: this excludes the description body of a pull request

.github/workflows/integer_benchmark.yml

@@ -0,0 +1,129 @@
+# Run integer benchmarks on an AWS instance and return parsed results to Slab CI bot.
+name: Integer benchmarks
+
+on:
+  workflow_dispatch:
+    inputs:
+      instance_id:
+        description: "Instance ID"
+        type: string
+      instance_image_id:
+        description: "Instance AMI ID"
+        type: string
+      instance_type:
+        description: "Instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: "Slab request ID"
+        type: string
+
+env:
+  CARGO_TERM_COLOR: always
+  RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  PARSE_INTEGER_BENCH_CSV_FILE: tfhe_rs_integer_benches_${{ github.sha }}.csv
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+
+jobs:
+  run-integer-benchmarks:
+    name: Execute integer benchmarks in EC2
+    runs-on: ${{ github.event.inputs.runner_name }}
+    if: ${{ !cancelled() }}
+    steps:
+      - name: Instance configuration used
+        run: |
+          echo "IDs: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+
+      - name: Get benchmark date
+        run: |
+          echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
+
+      - name: Checkout tfhe-rs repo with tags
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Set up home
+        # "Install rust" step require root user to have a HOME directory which is not set.
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install rust
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: nightly
+          override: true
+
+      - name: Run benchmarks with AVX512
+        run: |
+          make AVX512_SUPPORT=ON bench_integer
+
+      - name: Parse benchmarks to csv
+        run: |
+          make PARSE_INTEGER_BENCH_CSV_FILE=${{ env.PARSE_INTEGER_BENCH_CSV_FILE }} \
+            parse_integer_benches
+
+      - name: Upload csv results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_csv_integer
+          path: ${{ env.PARSE_INTEGER_BENCH_CSV_FILE }}
+
+      - name: Parse results
+        run: |
+          COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
+          COMMIT_HASH="$(git describe --tags --dirty)"
+          python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
+          --database tfhe_rs \
+          --hardware ${{ inputs.instance_type }} \
+          --project-version "${COMMIT_HASH}" \
+          --branch ${{ github.ref_name }} \
+          --commit-date "${COMMIT_DATE}" \
+          --bench-date "${{ env.BENCH_DATE }}" \
+          --walk-subdirs \
+          --name-suffix avx512 \
+          --throughput
+
+      - name: Upload parsed results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_integer
+          path: ${{ env.RESULTS_FILENAME }}
+
+      - name: Checkout Slab repo
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: zama-ai/slab
+          path: slab
+          token: ${{ secrets.CONCRETE_ACTIONS_TOKEN }}
+
+      - name: Send data to Slab
+        shell: bash
+        run: |
+          echo "Computing HMac on results file"
+          SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
+          echo "Sending results to Slab..."
+          curl -v -k \
+          -H "Content-Type: application/json" \
+          -H "X-Slab-Repository: ${{ github.repository }}" \
+          -H "X-Slab-Command: store_data_v2" \
+          -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
+          -d @${{ env.RESULTS_FILENAME }} \
+          ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Integer benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/integer_multi_bit_benchmark.yml

@@ -0,0 +1,129 @@
+# Run integer benchmarks with multi-bit cryptographic parameters on an AWS instance and return parsed results to Slab CI bot.
+name: Integer Multi-bit benchmarks
+
+on:
+  workflow_dispatch:
+    inputs:
+      instance_id:
+        description: "Instance ID"
+        type: string
+      instance_image_id:
+        description: "Instance AMI ID"
+        type: string
+      instance_type:
+        description: "Instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: "Slab request ID"
+        type: string
+
+env:
+  CARGO_TERM_COLOR: always
+  RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  PARSE_INTEGER_BENCH_CSV_FILE: tfhe_rs_integer_benches_${{ github.sha }}.csv
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+
+jobs:
+  run-integer-benchmarks:
+    name: Execute integer multi-bit benchmarks in EC2
+    runs-on: ${{ github.event.inputs.runner_name }}
+    if: ${{ !cancelled() }}
+    steps:
+      - name: Instance configuration used
+        run: |
+          echo "IDs: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+
+      - name: Get benchmark date
+        run: |
+          echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
+
+      - name: Checkout tfhe-rs repo with tags
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Set up home
+        # "Install rust" step require root user to have a HOME directory which is not set.
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install rust
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: nightly
+          override: true
+
+      - name: Run multi-bit benchmarks with AVX512
+        run: |
+          make AVX512_SUPPORT=ON bench_integer_multi_bit
+
+      - name: Parse benchmarks to csv
+        run: |
+          make PARSE_INTEGER_BENCH_CSV_FILE=${{ env.PARSE_INTEGER_BENCH_CSV_FILE }} \
+            parse_integer_benches
+
+      - name: Upload csv results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_csv_integer
+          path: ${{ env.PARSE_INTEGER_BENCH_CSV_FILE }}
+
+      - name: Parse results
+        run: |
+          COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
+          COMMIT_HASH="$(git describe --tags --dirty)"
+          python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
+          --database tfhe_rs \
+          --hardware ${{ inputs.instance_type }} \
+          --project-version "${COMMIT_HASH}" \
+          --branch ${{ github.ref_name }} \
+          --commit-date "${COMMIT_DATE}" \
+          --bench-date "${{ env.BENCH_DATE }}" \
+          --walk-subdirs \
+          --name-suffix avx512 \
+          --throughput
+
+      - name: Upload parsed results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_integer
+          path: ${{ env.RESULTS_FILENAME }}
+
+      - name: Checkout Slab repo
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: zama-ai/slab
+          path: slab
+          token: ${{ secrets.CONCRETE_ACTIONS_TOKEN }}
+
+      - name: Send data to Slab
+        shell: bash
+        run: |
+          echo "Computing HMac on results file"
+          SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
+          echo "Sending results to Slab..."
+          curl -v -k \
+          -H "Content-Type: application/json" \
+          -H "X-Slab-Repository: ${{ github.repository }}" \
+          -H "X-Slab-Command: store_data_v2" \
+          -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
+          -d @${{ env.RESULTS_FILENAME }} \
+          ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Integer benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/m1_tests.yml

@@ -4,11 +4,19 @@ on:
   workflow_dispatch:
   pull_request:
     types: [labeled]
+  # Have a nightly build for M1 tests
+  schedule:
+    # * is a special character in YAML so you have to quote this string
+    # At 22:00 every day
+    # Timezone is UTC, so Paris time is +2 during the summer and +1 during winter
+    - cron: "0 22 * * *"
 
 env:
   CARGO_TERM_COLOR: always
   RUSTFLAGS: "-C target-cpu=native"
   ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+  CARGO_PROFILE: release_lto_off
+  FAST_TESTS: "TRUE"
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref }}
@@ -16,11 +24,11 @@ concurrency:
 
 jobs:
   cargo-builds:
-    if: "github.event_name != 'pull_request' || contains(github.event.label.name, 'm1_test')"
+    if: ${{ (github.event_name == 'schedule' &&  github.repository == 'zama-ai/tfhe-rs') || github.event_name == 'workflow_dispatch' || contains(github.event.label.name, 'm1_test') }}
     runs-on: ["self-hosted", "m1mac"]
 
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+      - uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
 
       - name: Install latest stable
         uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
@@ -32,6 +40,10 @@ jobs:
         run: |
           make pcc
 
+      - name: Build Release core
+        run: |
+          make build_core
+
       - name: Build Release boolean
         run: |
           make build_boolean
@@ -40,9 +52,13 @@ jobs:
         run: |
           make build_shortint
 
-      - name: Build Release shortint and boolean
+      - name: Build Release integer
         run: |
-          make build_boolean_and_shortint
+          make build_integer
+
+      - name: Build Release tfhe full
+        run: |
+          make build_tfhe_full
 
       - name: Build Release c_api
         run: |
@@ -75,6 +91,23 @@ jobs:
         run: |
           make test_shortint_ci
 
+      - name: Run integer tests
+        run: |
+          make test_integer_ci
+
+      - name: Gen Keys if required
+        run: |
+          make GEN_KEY_CACHE_MULTI_BIT_ONLY=TRUE gen_key_cache
+
+      - name: Run shortint multi bit tests
+        run: |
+          make test_shortint_multi_bit_ci
+
+      # # These multi bit integer tests are too slow on M1 with low core count and low RAM
+      # - name: Run integer multi bit tests
+      #   run: |
+      #     make test_integer_multi_bit_ci
+
   remove_label:
     name: Remove m1_test label
     runs-on: ubuntu-latest
@@ -83,12 +116,13 @@ jobs:
     if: ${{ always() }}
     steps:
       - uses: actions-ecosystem/action-remove-labels@2ce5d41b4b6aa8503e285553f75ed56e0a40bae0
+        if: ${{ github.event_name == 'pull_request' }}
         with:
           labels: m1_test
           github_token: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Slack Notification
-        if: ${{ always() }}
+        if: ${{ needs.cargo-builds.result != 'skipped' }}
         continue-on-error: true
         uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
         env:

.github/workflows/make_release.yml

@@ -0,0 +1,84 @@
+# Publish new release of tfhe-rs on various platform.
+name: Publish release
+
+on:
+  workflow_dispatch:
+    inputs:
+      dry_run:
+        description: "Dry-run"
+        type: boolean
+        default: true
+      push_to_crates:
+        description: "Push to crate"
+        type: boolean
+        default: true
+      push_web_package:
+        description: "Push web js package"
+        type: boolean
+        default: true
+      push_node_package:
+        description: "Push node js package"
+        type: boolean
+        default: true
+
+env:
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+
+jobs:
+  publish_release:
+    name: Publish Release
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Publish crate.io package
+        if: ${{ inputs.push_to_crates }}
+        env:
+          CRATES_TOKEN: ${{ secrets.CARGO_REGISTRY_TOKEN }}
+          DRY_RUN: ${{ inputs.dry_run && '--dry-run' || '' }}
+        run: |
+          cargo publish -p tfhe --token ${{ env.CRATES_TOKEN }} ${{ env.DRY_RUN }}
+
+      - name: Build web package
+        if: ${{ inputs.push_web_package }}
+        run: |
+          make build_web_js_api
+
+      - name: Publish web package
+        if: ${{ inputs.push_web_package }}
+        uses: JS-DevTools/npm-publish@5a85faf05d2ade2d5b6682bfe5359915d5159c6c
+        with:
+          token: ${{ secrets.NPM_TOKEN }}
+          package: tfhe/pkg/package.json
+          dry-run: ${{ inputs.dry_run }}
+
+      - name: Build Node package
+        if: ${{ inputs.push_node_package }}
+        run: |
+          rm -rf tfhe/pkg
+
+          make build_node_js_api
+          sed -i 's/"tfhe"/"node-tfhe"/g' tfhe/pkg/package.json
+
+      - name: Publish Node package
+        if: ${{ inputs.push_node_package }}
+        uses: JS-DevTools/npm-publish@5a85faf05d2ade2d5b6682bfe5359915d5159c6c
+        with:
+          token: ${{ secrets.NPM_TOKEN }}
+          package: tfhe/pkg/package.json
+          dry-run: ${{ inputs.dry_run }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Integer benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/pbs_benchmark.yml

@@ -0,0 +1,117 @@
+# Run PBS benchmarks on an AWS instance and return parsed results to Slab CI bot.
+name: PBS benchmarks
+
+on:
+  workflow_dispatch:
+    inputs:
+      instance_id:
+        description: "Instance ID"
+        type: string
+      instance_image_id:
+        description: "Instance AMI ID"
+        type: string
+      instance_type:
+        description: "Instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: "Slab request ID"
+        type: string
+
+env:
+  CARGO_TERM_COLOR: always
+  RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+
+jobs:
+  run-pbs-benchmarks:
+    name: Execute PBS benchmarks in EC2
+    runs-on: ${{ github.event.inputs.runner_name }}
+    if: ${{ !cancelled() }}
+    steps:
+      - name: Instance configuration used
+        run: |
+          echo "IDs: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+
+      - name: Get benchmark date
+        run: |
+          echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
+
+      - name: Checkout tfhe-rs repo with tags
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Set up home
+        # "Install rust" step require root user to have a HOME directory which is not set.
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install rust
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: nightly
+          override: true
+
+      - name: Run benchmarks with AVX512
+        run: |
+          make AVX512_SUPPORT=ON bench_pbs
+
+      - name: Parse results
+        run: |
+          COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
+          COMMIT_HASH="$(git describe --tags --dirty)"
+          python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
+          --database tfhe_rs \
+          --hardware ${{ inputs.instance_type }} \
+          --project-version "${COMMIT_HASH}" \
+          --branch ${{ github.ref_name }} \
+          --commit-date "${COMMIT_DATE}" \
+          --bench-date "${{ env.BENCH_DATE }}" \
+          --name-suffix avx512 \
+          --walk-subdirs \
+          --throughput
+
+      - name: Upload parsed results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_pbs
+          path: ${{ env.RESULTS_FILENAME }}
+
+      - name: Checkout Slab repo
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: zama-ai/slab
+          path: slab
+          token: ${{ secrets.CONCRETE_ACTIONS_TOKEN }}
+
+      - name: Send data to Slab
+        shell: bash
+        run: |
+          echo "Computing HMac on downloaded artifact"
+          SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
+          echo "Sending results to Slab..."
+          curl -v -k \
+          -H "Content-Type: application/json" \
+          -H "X-Slab-Repository: ${{ github.repository }}" \
+          -H "X-Slab-Command: store_data_v2" \
+          -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
+          -d @${{ env.RESULTS_FILENAME }} \
+          ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "PBS benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/shortint_benchmark.yml

@@ -5,28 +5,25 @@ on:
   workflow_dispatch:
     inputs:
       instance_id:
-        description: 'Instance ID'
+        description: "Instance ID"
         type: string
       instance_image_id:
-        description: 'Instance AMI ID'
+        description: "Instance AMI ID"
         type: string
       instance_type:
-        description: 'Instance product type'
+        description: "Instance product type"
         type: string
       runner_name:
-        description: 'Action runner name'
+        description: "Action runner name"
         type: string
       request_id:
-        description: 'Slab request ID'
+        description: "Slab request ID"
         type: string
-      matrix_item:
-        description: 'Build matrix item'
-        type: string
-
 
 env:
   CARGO_TERM_COLOR: always
   RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
 
 jobs:
   run-shortint-benchmarks:
@@ -40,14 +37,13 @@ jobs:
           echo "AMI: ${{ inputs.instance_image_id }}"
           echo "Type: ${{ inputs.instance_type }}"
           echo "Request ID: ${{ inputs.request_id }}"
-          echo "Matrix item: ${{ inputs.matrix_item }}"
 
       - name: Get benchmark date
         run: |
           echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
 
       - name: Checkout tfhe-rs repo with tags
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           fetch-depth: 0
 
@@ -62,37 +58,24 @@ jobs:
           toolchain: nightly
           override: true
 
-      - name: Run benchmarks
+      - name: Run benchmarks with AVX512
         run: |
-          make bench_shortint
+          make AVX512_SUPPORT=ON bench_shortint
 
       - name: Parse results
         run: |
           COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
           COMMIT_HASH="$(git describe --tags --dirty)"
           python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
-          --database tfhe_rs_benchmarks \
+          --database tfhe_rs \
           --hardware ${{ inputs.instance_type }} \
           --project-version "${COMMIT_HASH}" \
           --branch ${{ github.ref_name }} \
           --commit-date "${COMMIT_DATE}" \
           --bench-date "${{ env.BENCH_DATE }}" \
-          --walk-subdirs
-
-      - name: Remove previous raw results
-        run: |
-          rm -rf target/criterion
-
-      - name: Run benchmarks with AVX512
-        run: |
-          make AVX512_SUPPORT=ON bench_shortint
-
-      - name: Parse AVX512 results
-        run: |
-          python3 ./ci/benchmark_parser.py target/criterion ${{ env.RESULTS_FILENAME }} \
           --walk-subdirs \
           --name-suffix avx512 \
-          --append-results
+          --throughput
 
       - name: Measure key sizes
         run: |
@@ -111,7 +94,7 @@ jobs:
           path: ${{ env.RESULTS_FILENAME }}
 
       - name: Checkout Slab repo
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           repository: zama-ai/slab
           path: slab
@@ -120,13 +103,25 @@ jobs:
       - name: Send data to Slab
         shell: bash
         run: |
-          echo "Computing HMac on downloaded artifact"
+          echo "Computing HMac on results file"
           SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
           echo "Sending results to Slab..."
           curl -v -k \
           -H "Content-Type: application/json" \
           -H "X-Slab-Repository: ${{ github.repository }}" \
-          -H "X-Slab-Command: store_data" \
+          -H "X-Slab-Command: store_data_v2" \
           -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
           -d @${{ env.RESULTS_FILENAME }} \
           ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "Shortint benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.github/workflows/start_benchmarks.yml

@@ -4,30 +4,103 @@ name: Start all benchmarks
 on:
   push:
     branches:
-      - 'main'
+      - "main"
   workflow_dispatch:
+    inputs:
+      # The input name must be the name of the slab command to launch
+      boolean_bench:
+        description: "Run Boolean benches"
+        type: boolean
+        default: true
+      shortint_bench:
+        description: "Run shortint benches"
+        type: boolean
+        default: true
+      integer_bench:
+        description: "Run integer benches"
+        type: boolean
+        default: true
+      integer_multi_bit_bench:
+        description: "Run integer multi bit benches"
+        type: boolean
+        default: true
+      pbs_bench:
+        description: "Run PBS benches"
+        type: boolean
+        default: true
+      wasm_client_bench:
+        description: "Run WASM client benches"
+        type: boolean
+        default: true
 
 jobs:
   start-benchmarks:
+    if: ${{ (github.event_name == 'push' && github.repository == 'zama-ai/tfhe-rs') || github.event_name == 'workflow_dispatch' }}
     strategy:
       matrix:
-        command: [boolean_bench, shortint_bench]
+        command: [boolean_bench, shortint_bench, integer_bench, integer_multi_bit_bench, pbs_bench, wasm_client_bench]
     runs-on: ubuntu-latest
     steps:
+      - name: Checkout tfhe-rs
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Check for file changes
+        id: changed-files
+        uses: tj-actions/changed-files@de0eba32790fb9bf87471b32855a30fc8f9d5fc6
+        with:
+          files_yaml: |
+            common_benches:
+              - toolchain.txt
+              - Makefile
+              - ci/slab.toml
+              - tfhe/Cargo.toml
+              - tfhe/src/core_crypto/**
+              - .github/workflows/start_benchmarks.yml
+            boolean_bench:
+              - tfhe/src/boolean/**
+              - tfhe/benches/boolean/**
+              - .github/workflows/boolean_benchmark.yml
+            shortint_bench:
+              - tfhe/src/shortint/**
+              - tfhe/benches/shortint/**
+              - .github/workflows/shortint_benchmark.yml
+            integer_bench:
+              - tfhe/src/shortint/**
+              - tfhe/src/integer/**
+              - tfhe/benches/integer/**
+              - .github/workflows/integer_benchmark.yml
+            integer_multi_bit_bench:
+              - tfhe/src/shortint/**
+              - tfhe/src/integer/**
+              - tfhe/benches/integer/**
+              - .github/workflows/integer_benchmark.yml
+            pbs_bench:
+              - tfhe/src/core_crypto/**
+              - tfhe/benches/core_crypto/**
+              - .github/workflows/pbs_benchmark.yml
+            wasm_client_bench:
+              - tfhe/web_wasm_parallel_tests/**
+              - .github/workflows/wasm_client_benchmark.yml
+
       - name: Checkout Slab repo
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           repository: zama-ai/slab
           path: slab
           token: ${{ secrets.CONCRETE_ACTIONS_TOKEN }}
 
       - name: Start AWS job in Slab
+        # If manually triggered check that the current bench has been requested
+        # Otherwise if it's on push check that files relevant to benchmarks have changed
+        if: (github.event_name == 'workflow_dispatch' && github.event.inputs[matrix.command] == 'true') || (github.event_name == 'push' && (steps.changed-files.outputs.common_benches_any_changed == 'true' || steps.changed-files.outputs[format('{0}_any_changed', matrix.command)] == 'true'))
         shell: bash
-        # TODO: step result must be correlated to HTTP return code.
         run: |
           echo -n '{"command": "${{ matrix.command }}", "git_ref": "${{ github.ref }}", "sha": "${{ github.sha }}"}' > command.json
           SIGNATURE="$(slab/scripts/hmac_calculator.sh command.json '${{ secrets.JOB_SECRET }}')"
           curl -v -k \
+          --fail-with-body \
           -H "Content-Type: application/json" \
           -H "X-Slab-Repository: ${{ github.repository }}" \
           -H "X-Slab-Command: start_aws" \

.github/workflows/sync_on_push.yml

@@ -13,7 +13,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repo
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
         with:
           fetch-depth: 0
       - name: Save repo
@@ -26,12 +26,12 @@ jobs:
         with:
           source_repo: "zama-ai/tfhe-rs"
           source_branch: "main"
-          destination_repo: ${{ secrets.SYNC_DEST_REPO }}
+          destination_repo: "https://${{ secrets.BOT_USERNAME }}:${{ secrets.CONCRETE_ACTIONS_TOKEN }}@github.com/${{ secrets.SYNC_DEST_REPO }}"
           destination_branch: "main"
       - name: git-sync tags
         uses: wei/git-sync@55c6b63b4f21607da0e9877ca9b4d11a29fc6d83
         with:
           source_repo: "zama-ai/tfhe-rs"
           source_branch: "refs/tags/*"
-          destination_repo: ${{ secrets.SYNC_DEST_REPO }}
+          destination_repo: "https://${{ secrets.BOT_USERNAME }}:${{ secrets.CONCRETE_ACTIONS_TOKEN }}@github.com/${{ secrets.SYNC_DEST_REPO }}"
           destination_branch: "refs/tags/*"

.github/workflows/trigger_aws_tests_on_pr.yml

@@ -3,15 +3,32 @@ name: PR AWS build trigger
 
 on:
   pull_request:
+  pull_request_review:
+    types: [submitted]
 
 jobs:
-  test:
+  trigger-tests:
     runs-on: ubuntu-latest
     permissions:
       pull-requests: write
     steps:
-      - uses: mshick/add-pr-comment@a65df5f64fc741e91c59b8359a4bc56e57aaf5b1
+      - name: Launch fast tests
+        if: ${{ github.event_name == 'pull_request' }}
+        uses: mshick/add-pr-comment@a65df5f64fc741e91c59b8359a4bc56e57aaf5b1
         with:
           allow-repeats: true
           message: |
+            @slab-ci cpu_fast_test
+
+      - name: Launch full tests suite
+        if: ${{ github.event_name == 'pull_request_review' && github.event.review.state == 'approved' }}
+        uses: mshick/add-pr-comment@a65df5f64fc741e91c59b8359a4bc56e57aaf5b1
+        with:
+          allow-repeats: true
+          message: |
+            Pull Request has been approved :tada:
+            Launching full test suite...
             @slab-ci cpu_test
+            @slab-ci cpu_integer_test
+            @slab-ci cpu_multi_bit_test
+            @slab-ci cpu_wasm_test

.github/workflows/wasm_client_benchmark.yml

@@ -0,0 +1,128 @@
+# Run WASM client benchmarks on an AWS instance and return parsed results to Slab CI bot.
+name: WASM client benchmarks
+
+on:
+  workflow_dispatch:
+    inputs:
+      instance_id:
+        description: "Instance ID"
+        type: string
+      instance_image_id:
+        description: "Instance AMI ID"
+        type: string
+      instance_type:
+        description: "Instance product type"
+        type: string
+      runner_name:
+        description: "Action runner name"
+        type: string
+      request_id:
+        description: "Slab request ID"
+        type: string
+
+env:
+  CARGO_TERM_COLOR: always
+  RESULTS_FILENAME: parsed_benchmark_results_${{ github.sha }}.json
+  ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+
+jobs:
+  run-wasm-client-benchmarks:
+    name: Execute WASM client benchmarks in EC2
+    runs-on: ${{ github.event.inputs.runner_name }}
+    if: ${{ !cancelled() }}
+    steps:
+      - name: Instance configuration used
+        run: |
+          echo "IDs: ${{ inputs.instance_id }}"
+          echo "AMI: ${{ inputs.instance_image_id }}"
+          echo "Type: ${{ inputs.instance_type }}"
+          echo "Request ID: ${{ inputs.request_id }}"
+
+      - name: Get benchmark date
+        run: |
+          echo "BENCH_DATE=$(date --iso-8601=seconds)" >> "${GITHUB_ENV}"
+
+      - name: Checkout tfhe-rs repo with tags
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          fetch-depth: 0
+
+      - name: Set up home
+        # "Install rust" step require root user to have a HOME directory which is not set.
+        run: |
+          echo "HOME=/home/ubuntu" >> "${GITHUB_ENV}"
+
+      - name: Install rust
+        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af
+        with:
+          toolchain: nightly
+          override: true
+
+      - name: Run benchmarks
+        run: |
+          make install_node
+          make ci_bench_web_js_api_parallel
+
+      - name: Parse results
+        run: |
+          make parse_wasm_benchmarks
+
+          COMMIT_DATE="$(git --no-pager show -s --format=%cd --date=iso8601-strict ${{ github.sha }})"
+          COMMIT_HASH="$(git describe --tags --dirty)"
+          python3 ./ci/benchmark_parser.py tfhe/wasm_pk_gen.csv ${{ env.RESULTS_FILENAME }} \
+          --database tfhe_rs \
+          --hardware ${{ inputs.instance_type }} \
+          --project-version "${COMMIT_HASH}" \
+          --branch ${{ github.ref_name }} \
+          --commit-date "${COMMIT_DATE}" \
+          --bench-date "${{ env.BENCH_DATE }}" \
+          --key-gen
+
+      - name: Measure public key and ciphertext sizes in HL Api
+        run: |
+          make measure_hlapi_compact_pk_ct_sizes
+
+      - name: Parse key and ciphertext sizes results
+        run: |
+          python3 ./ci/benchmark_parser.py tfhe/hlapi_cpk_and_cctl_sizes.csv ${{ env.RESULTS_FILENAME }} \
+          --key-gen \
+          --append-results
+
+      - name: Upload parsed results artifact
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce
+        with:
+          name: ${{ github.sha }}_wasm
+          path: ${{ env.RESULTS_FILENAME }}
+
+      - name: Checkout Slab repo
+        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9
+        with:
+          repository: zama-ai/slab
+          path: slab
+          token: ${{ secrets.CONCRETE_ACTIONS_TOKEN }}
+
+      - name: Send data to Slab
+        shell: bash
+        run: |
+          echo "Computing HMac on results file"
+          SIGNATURE="$(slab/scripts/hmac_calculator.sh ${{ env.RESULTS_FILENAME }} '${{ secrets.JOB_SECRET }}')"
+          echo "Sending results to Slab..."
+          curl -v -k \
+          -H "Content-Type: application/json" \
+          -H "X-Slab-Repository: ${{ github.repository }}" \
+          -H "X-Slab-Command: store_data_v2" \
+          -H "X-Hub-Signature-256: sha256=${SIGNATURE}" \
+          -d @${{ env.RESULTS_FILENAME }} \
+          ${{ secrets.SLAB_URL }}
+
+      - name: Slack Notification
+        if: ${{ failure() }}
+        continue-on-error: true
+        uses: rtCamp/action-slack-notify@12e36fc18b0689399306c2e0b3e0f2978b7f1ee7
+        env:
+          SLACK_COLOR: ${{ job.status }}
+          SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }}
+          SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png
+          SLACK_MESSAGE: "WASM benchmarks failed. (${{ env.ACTION_RUN_URL }})"
+          SLACK_USERNAME: ${{ secrets.BOT_USERNAME }}
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}

.gitignore

@@ -3,7 +3,13 @@ target/
 .vscode/
 
 # Path we use for internal-keycache during tests
-keys/
+./keys/
+# In case of symlinked keys
+./keys
 
 **/Cargo.lock
 **/*.bin
+
+# Some of our bench outputs
+/tfhe/benchmarks_parameters
+**/*.csv

CODE_OF_CONDUCT.md

@@ -0,0 +1,131 @@
+# Contributor Covenant Code of Conduct
+
+## Our pledge
+
+We as members, contributors, and leaders pledge to make participation in our
+community a harassment-free experience for everyone, regardless of age, body
+size, visible or invisible disability, ethnicity, sex characteristics, gender
+identity and expression, level of experience, education, socio-economic status,
+nationality, personal appearance, race, caste, color, religion, or sexual
+identity and orientation.
+
+We pledge to act and interact in ways that contribute to an open, welcoming,
+diverse, inclusive, and healthy community.
+
+## Our standards
+
+Examples of behavior that contributes to a positive environment for our
+community include:
+
+- Demonstrating empathy and kindness toward other people
+- Being respectful of differing opinions, viewpoints, and experiences
+- Giving and gracefully accepting constructive feedback
+- Accepting responsibility and apologizing to those affected by our mistakes,
+  and learning from the experience
+- Focusing on what is best not just for us as individuals, but for the overall
+  community
+
+Examples of unacceptable behavior include:
+
+- The use of sexualized language or imagery, and sexual attention or advances of
+  any kind
+- Trolling, insulting or derogatory comments, and personal or political attacks
+- Public or private harassment
+- Publishing others' private information, such as a physical or email address,
+  without their explicit permission
+- Other conduct which could reasonably be considered inappropriate in a
+  professional setting
+
+## Enforcement responsibilities
+
+Community leaders are responsible for clarifying and enforcing our standards of
+acceptable behavior and will take appropriate and fair corrective action in
+response to any behavior that they deem inappropriate, threatening, offensive,
+or harmful.
+
+Community leaders have the right and responsibility to remove, edit, or reject
+comments, commits, code, wiki edits, issues, and other contributions that are
+not aligned to this Code of Conduct, and will communicate reasons for moderation
+decisions when appropriate.
+
+## Scope
+
+This Code of Conduct applies within all community spaces, and also applies when
+an individual is officially representing the community in public spaces.
+Examples of representing our community include using an official e-mail address,
+posting via an official social media account, or acting as an appointed
+representative at an online or offline event.
+
+## Enforcement
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may be
+reported by contacting us anonymously through [this form](https://forms.gle/569j3cZqGRFgrR3u9).
+All complaints will be reviewed and investigated promptly and fairly.
+
+All community leaders are obligated to respect the privacy and security of the
+reporter of any incident.
+
+## Enforcement guidelines
+
+Community leaders will follow these Community Impact Guidelines in determining
+the consequences for any action they deem in violation of this Code of Conduct:
+
+### 1. Correction
+
+**Community Impact**: Use of inappropriate language or other behavior deemed
+unprofessional or unwelcome in the community.
+
+**Consequence**: A private, written warning from community leaders, providing
+clarity around the nature of the violation and an explanation of why the
+behavior was inappropriate. A public apology may be requested.
+
+### 2. Warning
+
+**Community Impact**: A violation through a single incident or series of
+actions.
+
+**Consequence**: A warning with consequences for continued behavior. No
+interaction with the people involved, including unsolicited interaction with
+those enforcing the Code of Conduct, for a specified period of time. This
+includes avoiding interactions in community spaces as well as external channels
+like social media. Violating these terms may lead to a temporary or permanent
+ban.
+
+### 3. Temporary ban
+
+**Community Impact**: A serious violation of community standards, including
+sustained inappropriate behavior.
+
+**Consequence**: A temporary ban from any sort of interaction or public
+communication with the community for a specified period of time. No public or
+private interaction with the people involved, including unsolicited interaction
+with those enforcing the Code of Conduct, is allowed during this period.
+Violating these terms may lead to a permanent ban.
+
+### 4. Permanent ban
+
+**Community Impact**: Demonstrating a pattern of violation of community
+standards, including sustained inappropriate behavior, harassment of an
+individual, or aggression toward or disparagement of classes of individuals.
+
+**Consequence**: A permanent ban from any sort of public interaction within the
+community.
+
+## Attribution
+
+This Code of Conduct is adapted from the [Contributor Covenant][homepage],
+version 2.1, available at
+[https://www.contributor-covenant.org/version/2/1/code_of_conduct.html][v2.1].
+
+Community Impact Guidelines were inspired by
+[Mozilla's code of conduct enforcement ladder][mozilla coc].
+
+For answers to common questions about this code of conduct, see the FAQ at
+[https://www.contributor-covenant.org/faq][faq]. Translations are available at
+[https://www.contributor-covenant.org/translations][translations].
+
+[faq]: https://www.contributor-covenant.org/faq
+[homepage]: https://www.contributor-covenant.org
+[mozilla coc]: https://github.com/mozilla/diversity
+[translations]: https://www.contributor-covenant.org/translations
+[v2.1]: https://www.contributor-covenant.org/version/2/1/code_of_conduct.html

Cargo.toml

@@ -1,9 +1,19 @@
 [workspace]
 resolver = "2"
-members = ["tfhe", "tasks"]
+members = ["tfhe", "tasks", "apps/trivium"]
 
 [profile.bench]
 lto = "fat"
 
 [profile.release]
 lto = "fat"
+
+[profile.release_lto_off]
+inherits = "release"
+lto = "off"
+
+# Compiles much faster for tests and allows reasonable performance for iterating
+[profile.devo]
+inherits = "dev"
+opt-level = 3
+lto = "off"

LICENSE

@@ -1,28 +1,28 @@
-BSD 3-Clause Clear License
-
-Copyright © 2022 ZAMA.
-All rights reserved.
-
-Redistribution and use in source and binary forms, with or without modification,
-are permitted provided that the following conditions are met:
-
-1. Redistributions of source code must retain the above copyright notice, this
-list of conditions and the following disclaimer.
-
-2. Redistributions in binary form must reproduce the above copyright notice, this
-list of conditions and the following disclaimer in the documentation and/or other
-materials provided with the distribution.
-
-3. Neither the name of ZAMA nor the names of its contributors may be used to endorse
-or promote products derived from this software without specific prior written permission.
-
-NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED BY THIS LICENSE.
-THIS SOFTWARE IS PROVIDED BY THE ZAMA AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
-IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
-MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
-ZAMA OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
-OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
-OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
-ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
-NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
-ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+BSD 3-Clause Clear License
+
+Copyright © 2023 ZAMA.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without modification,
+are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this
+list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this
+list of conditions and the following disclaimer in the documentation and/or other
+materials provided with the distribution.
+
+3. Neither the name of ZAMA nor the names of its contributors may be used to endorse
+or promote products derived from this software without specific prior written permission.
+
+NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED BY THIS LICENSE.
+THIS SOFTWARE IS PROVIDED BY THE ZAMA AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
+IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
+ZAMA OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
+OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Makefile

@@ -1,13 +1,20 @@
 SHELL:=$(shell /usr/bin/env which bash)
+OS:=$(shell uname)
 RS_CHECK_TOOLCHAIN:=$(shell cat toolchain.txt | tr -d '\n')
 CARGO_RS_CHECK_TOOLCHAIN:=+$(RS_CHECK_TOOLCHAIN)
 TARGET_ARCH_FEATURE:=$(shell ./scripts/get_arch_feature.sh)
 RS_BUILD_TOOLCHAIN:=$(shell \
 	( (echo $(TARGET_ARCH_FEATURE) | grep -q x86) && echo stable) || echo $(RS_CHECK_TOOLCHAIN))
 CARGO_RS_BUILD_TOOLCHAIN:=+$(RS_BUILD_TOOLCHAIN)
-MIN_RUST_VERSION:=1.65
+CARGO_PROFILE?=release
+MIN_RUST_VERSION:=$(shell grep rust-version tfhe/Cargo.toml | cut -d '=' -f 2 | xargs)
 AVX512_SUPPORT?=OFF
 WASM_RUSTFLAGS:=
+BIG_TESTS_INSTANCE?=FALSE
+GEN_KEY_CACHE_MULTI_BIT_ONLY?=FALSE
+PARSE_INTEGER_BENCH_CSV_FILE?=tfhe_rs_integer_benches.csv
+FAST_TESTS?=FALSE
+BENCH_OP_FLAVOR?=DEFAULT
 # This is done to avoid forgetting it, we still precise the RUSTFLAGS in the commands to be able to
 # copy paste the command in the terminal and change them if required without forgetting the flags
 export RUSTFLAGS?=-C target-cpu=native
@@ -18,6 +25,16 @@ else
 		AVX512_FEATURE=
 endif
 
+ifeq ($(GEN_KEY_CACHE_MULTI_BIT_ONLY),TRUE)
+		MULTI_BIT_ONLY=--multi-bit-only
+else
+		MULTI_BIT_ONLY=
+endif
+
+# Variables used only for regex_engine example
+REGEX_STRING?=''
+REGEX_PATTERN?=''
+
 .PHONY: rs_check_toolchain # Echo the rust toolchain used for checks
 rs_check_toolchain:
 	@echo $(RS_CHECK_TOOLCHAIN)
@@ -49,6 +66,19 @@ install_cargo_nextest: install_rs_build_toolchain
 	cargo $(CARGO_RS_BUILD_TOOLCHAIN) install cargo-nextest --locked || \
 	( echo "Unable to install cargo nextest, unknown error." && exit 1 )
 
+.PHONY: install_wasm_pack # Install wasm-pack to build JS packages
+install_wasm_pack: install_rs_build_toolchain
+	@wasm-pack --version > /dev/null 2>&1 || \
+	cargo $(CARGO_RS_BUILD_TOOLCHAIN) install wasm-pack || \
+	( echo "Unable to install cargo wasm-pack, unknown error." && exit 1 )
+
+.PHONY: install_node # Install last version of NodeJS via nvm
+install_node:
+	curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.3/install.sh | $(SHELL)
+	source ~/.bashrc
+	$(SHELL) -i -c 'nvm install node' || \
+	( echo "Unable to install node, unknown error." && exit 1 )
+
 .PHONY: fmt # Format rust code
 fmt: install_rs_check_toolchain
 	cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" fmt
@@ -57,6 +87,15 @@ fmt: install_rs_check_toolchain
 check_fmt: install_rs_check_toolchain
 	cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" fmt --check
 
+.PHONY: clippy_core # Run clippy lints on core_crypto with and without experimental features
+clippy_core: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy \
+		--features=$(TARGET_ARCH_FEATURE) \
+		-p tfhe -- --no-deps -D warnings
+	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy \
+		--features=$(TARGET_ARCH_FEATURE),experimental \
+		-p tfhe -- --no-deps -D warnings
+
 .PHONY: clippy_boolean # Run clippy lints enabling the boolean features
 clippy_boolean: install_rs_check_toolchain
 	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy \
@@ -69,10 +108,16 @@ clippy_shortint: install_rs_check_toolchain
 		--features=$(TARGET_ARCH_FEATURE),shortint \
 		-p tfhe -- --no-deps -D warnings
 
-.PHONY: clippy # Run clippy lints enabling the boolean, shortint
-clippy: install_rs_check_toolchain
+.PHONY: clippy_integer # Run clippy lints enabling the integer features
+clippy_integer: install_rs_check_toolchain
 	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy \
-		--features=$(TARGET_ARCH_FEATURE),boolean,shortint \
+		--features=$(TARGET_ARCH_FEATURE),integer \
+		-p tfhe -- --no-deps -D warnings
+
+.PHONY: clippy # Run clippy lints enabling the boolean, shortint, integer
+clippy: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy --all-targets \
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer \
 		-p tfhe -- --no-deps -D warnings
 
 .PHONY: clippy_c_api # Run clippy lints enabling the boolean, shortint and the C API
@@ -81,10 +126,10 @@ clippy_c_api: install_rs_check_toolchain
 		--features=$(TARGET_ARCH_FEATURE),boolean-c-api,shortint-c-api \
 		-p tfhe -- --no-deps -D warnings
 
-.PHONY: clippy_js_wasm_api # Run clippy lints enabling the boolean, shortint and the js wasm API
+.PHONY: clippy_js_wasm_api # Run clippy lints enabling the boolean, shortint, integer and the js wasm API
 clippy_js_wasm_api: install_rs_check_toolchain
 	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy \
-		--features=boolean-client-js-wasm-api,shortint-client-js-wasm-api \
+		--features=boolean-client-js-wasm-api,shortint-client-js-wasm-api,integer-client-js-wasm-api \
 		-p tfhe -- --no-deps -D warnings
 
 .PHONY: clippy_tasks # Run clippy lints on helper tasks crate.
@@ -95,87 +140,209 @@ clippy_tasks:
 .PHONY: clippy_all_targets # Run clippy lints on all targets (benches, examples, etc.)
 clippy_all_targets:
 	RUSTFLAGS="$(RUSTFLAGS)" cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" clippy --all-targets \
-		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,internal-keycache \
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer,internal-keycache \
 		-p tfhe -- --no-deps -D warnings
 
 .PHONY: clippy_all # Run all clippy targets
-clippy_all: clippy clippy_boolean clippy_shortint clippy_all_targets clippy_c_api \
-clippy_js_wasm_api clippy_tasks
+clippy_all: clippy clippy_boolean clippy_shortint clippy_integer clippy_all_targets clippy_c_api \
+clippy_js_wasm_api clippy_tasks clippy_core
+
+.PHONY: clippy_fast # Run main clippy targets
+clippy_fast: clippy clippy_all_targets clippy_c_api clippy_js_wasm_api clippy_tasks clippy_core
 
 .PHONY: gen_key_cache # Run the script to generate keys and cache them for shortint tests
 gen_key_cache: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) run --release \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
 		--example generates_test_keys \
-		--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache -p tfhe
+		--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache -p tfhe -- \
+		$(MULTI_BIT_ONLY)
+
+.PHONY: build_core # Build core_crypto without experimental features
+build_core: install_rs_build_toolchain install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE) -p tfhe
+	@if [[ "$(AVX512_SUPPORT)" == "ON" ]]; then \
+		RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+			--features=$(TARGET_ARCH_FEATURE),$(AVX512_FEATURE) -p tfhe; \
+	fi
+
+.PHONY: build_core_experimental # Build core_crypto with experimental features
+build_core_experimental: install_rs_build_toolchain install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),experimental -p tfhe
+	@if [[ "$(AVX512_SUPPORT)" == "ON" ]]; then \
+		RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+			--features=$(TARGET_ARCH_FEATURE),experimental,$(AVX512_FEATURE) -p tfhe; \
+	fi
 
 .PHONY: build_boolean # Build with boolean enabled
 build_boolean: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --release \
-		--features=$(TARGET_ARCH_FEATURE),boolean -p tfhe
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean -p tfhe --all-targets
 
 .PHONY: build_shortint # Build with shortint enabled
 build_shortint: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --release \
-		--features=$(TARGET_ARCH_FEATURE),shortint -p tfhe
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),shortint -p tfhe --all-targets
 
-.PHONY: build_boolean_and_shortint # Build with boolean and shortint enabled
-build_boolean_and_shortint: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --release \
-		--features=$(TARGET_ARCH_FEATURE),boolean,shortint -p tfhe
+.PHONY: build_integer # Build with integer enabled
+build_integer: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),integer -p tfhe --all-targets
 
-.PHONY: build_c_api # Build the C API for boolean and shortint
-build_c_api: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --release \
-		--features=$(TARGET_ARCH_FEATURE),boolean-c-api,shortint-c-api -p tfhe
+.PHONY: build_tfhe_full # Build with boolean, shortint and integer enabled
+build_tfhe_full: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer -p tfhe --all-targets
+
+.PHONY: build_c_api # Build the C API for boolean, shortint and integer
+build_c_api: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean-c-api,shortint-c-api,high-level-c-api \
+		-p tfhe
+
+.PHONY: build_c_api_experimental_deterministic_fft # Build the C API for boolean, shortint and integer with experimental deterministic FFT
+build_c_api_experimental_deterministic_fft: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) build --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean-c-api,shortint-c-api,high-level-c-api,experimental-force_fft_algo_dif4 \
+		-p tfhe
 
 .PHONY: build_web_js_api # Build the js API targeting the web browser
-build_web_js_api: install_rs_build_toolchain
+build_web_js_api: install_rs_build_toolchain install_wasm_pack
 	cd tfhe && \
 	RUSTFLAGS="$(WASM_RUSTFLAGS)" rustup run "$(RS_BUILD_TOOLCHAIN)" \
 		wasm-pack build --release --target=web \
-		--features=boolean-client-js-wasm-api,shortint-client-js-wasm-api
+		-- --features=boolean-client-js-wasm-api,shortint-client-js-wasm-api,integer-client-js-wasm-api
+
+.PHONY: build_web_js_api_parallel # Build the js API targeting the web browser with parallelism support
+build_web_js_api_parallel: install_rs_check_toolchain install_wasm_pack
+	cd tfhe && \
+	rustup component add rust-src --toolchain $(RS_CHECK_TOOLCHAIN) && \
+	RUSTFLAGS="$(WASM_RUSTFLAGS) -C target-feature=+atomics,+bulk-memory,+mutable-globals" rustup run $(RS_CHECK_TOOLCHAIN) \
+		wasm-pack build --release --target=web \
+		-- --features=boolean-client-js-wasm-api,shortint-client-js-wasm-api,integer-client-js-wasm-api,parallel-wasm-api \
+		-Z build-std=panic_abort,std
 
 .PHONY: build_node_js_api # Build the js API targeting nodejs
-build_node_js_api: install_rs_build_toolchain
+build_node_js_api: install_rs_build_toolchain install_wasm_pack
 	cd tfhe && \
 	RUSTFLAGS="$(WASM_RUSTFLAGS)" rustup run "$(RS_BUILD_TOOLCHAIN)" \
 		wasm-pack build --release --target=nodejs \
-		--features=boolean-client-js-wasm-api,shortint-client-js-wasm-api
+		-- --features=boolean-client-js-wasm-api,shortint-client-js-wasm-api,integer-client-js-wasm-api
 
-.PHONY: test_core_crypto # Run the tests of the core_crypto module
-test_core_crypto: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --release \
-		--features=$(TARGET_ARCH_FEATURE) -p tfhe -- core_crypto::
+.PHONY: test_core_crypto # Run the tests of the core_crypto module including experimental ones
+test_core_crypto: install_rs_build_toolchain install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),experimental -p tfhe -- core_crypto::
+	@if [[ "$(AVX512_SUPPORT)" == "ON" ]]; then \
+		RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+			--features=$(TARGET_ARCH_FEATURE),experimental,$(AVX512_FEATURE) -p tfhe -- core_crypto::; \
+	fi
 
 .PHONY: test_boolean # Run the tests of the boolean module
 test_boolean: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --release \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
 		--features=$(TARGET_ARCH_FEATURE),boolean -p tfhe -- boolean::
 
-.PHONY: test_c_api # Run the tests for the C API
-test_c_api: build_c_api
+.PHONY: test_c_api_rs # Run the rust tests for the C API
+test_c_api_rs: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean-c-api,shortint-c-api,high-level-c-api \
+		-p tfhe \
+		c_api
+
+.PHONY: test_c_api_c # Run the C tests for the C API
+test_c_api_c: build_c_api
 	./scripts/c_api_tests.sh
 
+.PHONY: test_c_api # Run all the tests for the C API
+test_c_api: test_c_api_rs test_c_api_c
+
 .PHONY: test_shortint_ci # Run the tests for shortint ci
 test_shortint_ci: install_rs_build_toolchain install_cargo_nextest
-	./scripts/shortint-tests.sh $(CARGO_RS_BUILD_TOOLCHAIN)
+	BIG_TESTS_INSTANCE="$(BIG_TESTS_INSTANCE)" \
+	FAST_TESTS="$(FAST_TESTS)" \
+		./scripts/shortint-tests.sh --rust-toolchain $(CARGO_RS_BUILD_TOOLCHAIN) \
+		--cargo-profile "$(CARGO_PROFILE)"
+
+.PHONY: test_shortint_multi_bit_ci # Run the tests for shortint ci running only multibit tests
+test_shortint_multi_bit_ci: install_rs_build_toolchain install_cargo_nextest
+	BIG_TESTS_INSTANCE="$(BIG_TESTS_INSTANCE)" \
+	FAST_TESTS="$(FAST_TESTS)" \
+		./scripts/shortint-tests.sh --rust-toolchain $(CARGO_RS_BUILD_TOOLCHAIN) \
+		--cargo-profile "$(CARGO_PROFILE)" --multi-bit
 
 .PHONY: test_shortint # Run all the tests for shortint
 test_shortint: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --release \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
 		--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache -p tfhe -- shortint::
 
+.PHONY: test_integer_ci # Run the tests for integer ci
+test_integer_ci: install_rs_build_toolchain install_cargo_nextest
+	BIG_TESTS_INSTANCE="$(BIG_TESTS_INSTANCE)" \
+	FAST_TESTS="$(FAST_TESTS)" \
+		./scripts/integer-tests.sh --rust-toolchain $(CARGO_RS_BUILD_TOOLCHAIN) \
+		--cargo-profile "$(CARGO_PROFILE)"
+
+.PHONY: test_integer_multi_bit_ci # Run the tests for integer ci running only multibit tests
+test_integer_multi_bit_ci: install_rs_build_toolchain install_cargo_nextest
+	BIG_TESTS_INSTANCE="$(BIG_TESTS_INSTANCE)" \
+	FAST_TESTS="$(FAST_TESTS)" \
+		./scripts/integer-tests.sh --rust-toolchain $(CARGO_RS_BUILD_TOOLCHAIN) \
+		--cargo-profile "$(CARGO_PROFILE)" --multi-bit
+
+.PHONY: test_integer # Run all the tests for integer
+test_integer: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),integer,internal-keycache -p tfhe -- integer::
+
+.PHONY: test_high_level_api # Run all the tests for high_level_api
+test_high_level_api: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer,internal-keycache -p tfhe \
+		-- high_level_api::
+
 .PHONY: test_user_doc # Run tests from the .md documentation
 test_user_doc: install_rs_build_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --release --doc \
-		--features=$(TARGET_ARCH_FEATURE),shortint,boolean,internal-keycache -p tfhe \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) --doc \
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer,internal-keycache -p tfhe \
 		-- test_user_docs::
 
+.PHONY: test_regex_engine # Run tests for regex_engine example
+test_regex_engine: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--example regex_engine \
+		--features=$(TARGET_ARCH_FEATURE),integer
+
+.PHONY: test_sha256_bool # Run tests for sha256_bool example
+test_sha256_bool: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		--example sha256_bool \
+		--features=$(TARGET_ARCH_FEATURE),boolean
+
+.PHONY: test_examples # Run tests for examples
+test_examples: test_sha256_bool test_regex_engine
+
+.PHONY: test_trivium # Run tests for trivium
+test_trivium: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		trivium --features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer \
+		-- --test-threads=1
+
+.PHONY: test_kreyvium # Run tests for kreyvium
+test_kreyvium: install_rs_build_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --profile $(CARGO_PROFILE) \
+		kreyvium --features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer \
+		-- --test-threads=1
+
 .PHONY: doc # Build rust doc
 doc: install_rs_check_toolchain
 	RUSTDOCFLAGS="--html-in-header katex-header.html -Dwarnings" \
 	cargo "$(CARGO_RS_CHECK_TOOLCHAIN)" doc \
-		--features=$(TARGET_ARCH_FEATURE),boolean,shortint --no-deps
+		--features=$(TARGET_ARCH_FEATURE),boolean,shortint,integer --no-deps
+
+.PHONY: docs # Build rust doc alias for doc
+docs: doc
 
 .PHONY: format_doc_latex # Format the documentation latex equations to avoid broken rendering.
 format_doc_latex:
@@ -187,10 +354,15 @@ format_doc_latex:
 	@printf "\n===============================\n"
 
 .PHONY: check_compile_tests # Build tests in debug without running them
-check_compile_tests: build_c_api
+check_compile_tests:
 	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_BUILD_TOOLCHAIN) test --no-run \
-		--features=$(TARGET_ARCH_FEATURE),shortint,boolean,internal-keycache -p tfhe && \
-		./scripts/c_api_tests.sh --build-only
+		--features=$(TARGET_ARCH_FEATURE),experimental,boolean,shortint,integer,internal-keycache \
+		-p tfhe
+
+	@if [[ "$(OS)" == "Linux" || "$(OS)" == "Darwin" ]]; then \
+		"$(MAKE)" build_c_api; \
+		./scripts/c_api_tests.sh --build-only; \
+	fi
 
 .PHONY: build_nodejs_test_docker # Build a docker image with tools to run nodejs tests for wasm API
 build_nodejs_test_docker:
@@ -212,13 +384,46 @@ test_nodejs_wasm_api_in_docker: build_nodejs_test_docker
 test_nodejs_wasm_api: build_node_js_api
 	cd tfhe && node --test js_on_wasm_tests
 
+.PHONY: test_web_js_api_parallel # Run tests for the web wasm api
+test_web_js_api_parallel: build_web_js_api_parallel
+	$(MAKE) -C tfhe/web_wasm_parallel_tests test
+
+.PHONY: ci_test_web_js_api_parallel # Run tests for the web wasm api
+ci_test_web_js_api_parallel: build_web_js_api_parallel
+	source ~/.nvm/nvm.sh && \
+	nvm use node && \
+	$(MAKE) -C tfhe/web_wasm_parallel_tests test-ci
+
 .PHONY: no_tfhe_typo # Check we did not invert the h and f in tfhe
 no_tfhe_typo:
 	@./scripts/no_tfhe_typo.sh
 
+.PHONY: no_dbg_log # Check we did not leave dbg macro calls in the rust code
+no_dbg_log:
+	@./scripts/no_dbg_calls.sh
+
+#
+# Benchmarks
+#
+
+.PHONY: bench_integer # Run benchmarks for integer
+bench_integer: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" __TFHE_RS_BENCH_OP_FLAVOR=$(BENCH_OP_FLAVOR) \
+	cargo $(CARGO_RS_CHECK_TOOLCHAIN) bench \
+	--bench integer-bench \
+	--features=$(TARGET_ARCH_FEATURE),integer,internal-keycache,$(AVX512_FEATURE) -p tfhe --
+
+.PHONY: bench_integer_multi_bit # Run benchmarks for integer using multi-bit parameters
+bench_integer_multi_bit: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" __TFHE_RS_BENCH_TYPE=MULTI_BIT __TFHE_RS_BENCH_OP_FLAVOR=$(BENCH_OP_FLAVOR) \
+	cargo $(CARGO_RS_CHECK_TOOLCHAIN) bench \
+	--bench integer-bench \
+	--features=$(TARGET_ARCH_FEATURE),integer,internal-keycache,$(AVX512_FEATURE) -p tfhe --
+
 .PHONY: bench_shortint # Run benchmarks for shortint
 bench_shortint: install_rs_check_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) bench \
+	RUSTFLAGS="$(RUSTFLAGS)" __TFHE_RS_BENCH_OP_FLAVOR=$(BENCH_OP_FLAVOR) \
+	cargo $(CARGO_RS_CHECK_TOOLCHAIN) bench \
 	--bench shortint-bench \
 	--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache,$(AVX512_FEATURE) -p tfhe
 
@@ -228,20 +433,86 @@ bench_boolean: install_rs_check_toolchain
 	--bench boolean-bench \
 	--features=$(TARGET_ARCH_FEATURE),boolean,internal-keycache,$(AVX512_FEATURE) -p tfhe
 
+.PHONY: bench_pbs # Run benchmarks for PBS
+bench_pbs: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) bench \
+	--bench pbs-bench \
+	--features=$(TARGET_ARCH_FEATURE),boolean,shortint,internal-keycache,$(AVX512_FEATURE) -p tfhe
+
+.PHONY: bench_web_js_api_parallel # Run benchmarks for the web wasm api
+bench_web_js_api_parallel: build_web_js_api_parallel
+	$(MAKE) -C tfhe/web_wasm_parallel_tests bench
+
+.PHONY: ci_bench_web_js_api_parallel # Run benchmarks for the web wasm api
+ci_bench_web_js_api_parallel: build_web_js_api_parallel
+	source ~/.nvm/nvm.sh && \
+	nvm use node && \
+	$(MAKE) -C tfhe/web_wasm_parallel_tests bench-ci
+
+#
+# Utility tools
+#
+
+.PHONY: measure_hlapi_compact_pk_ct_sizes # Measure sizes of public keys and ciphertext for high-level API
+measure_hlapi_compact_pk_ct_sizes: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
+	--example hlapi_compact_pk_ct_sizes \
+	--features=$(TARGET_ARCH_FEATURE),integer,internal-keycache
+
 .PHONY: measure_shortint_key_sizes # Measure sizes of bootstrapping and key switching keys for shortint
 measure_shortint_key_sizes: install_rs_check_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
 	--example shortint_key_sizes \
 	--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache
 
 .PHONY: measure_boolean_key_sizes # Measure sizes of bootstrapping and key switching keys for boolean
 measure_boolean_key_sizes: install_rs_check_toolchain
-	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run \
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
 	--example boolean_key_sizes \
 	--features=$(TARGET_ARCH_FEATURE),boolean,internal-keycache
 
+.PHONY: parse_integer_benches # Run python parser to output a csv containing integer benches data
+parse_integer_benches:
+	python3 ./ci/parse_integer_benches_to_csv.py \
+		--criterion-dir target/criterion \
+		--output-file "$(PARSE_INTEGER_BENCH_CSV_FILE)"
+
+.PHONY: parse_wasm_benchmarks # Parse benchmarks performed with WASM web client into a CSV file
+parse_wasm_benchmarks: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
+	--example wasm_benchmarks_parser \
+	--features=$(TARGET_ARCH_FEATURE),shortint,internal-keycache \
+	-- web_wasm_parallel_tests/test/benchmark_results
+
+#
+# Real use case examples
+#
+
+.PHONY: regex_engine # Run regex_engine example
+regex_engine: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
+	--example regex_engine \
+	--features=$(TARGET_ARCH_FEATURE),integer \
+	-- $(REGEX_STRING) $(REGEX_PATTERN)
+
+.PHONY: dark_market # Run dark market example
+dark_market: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
+	--example dark_market \
+	--features=$(TARGET_ARCH_FEATURE),integer,internal-keycache \
+	-- fhe-modified fhe-parallel plain fhe
+
+.PHONY: sha256_bool # Run sha256_bool example
+sha256_bool: install_rs_check_toolchain
+	RUSTFLAGS="$(RUSTFLAGS)" cargo $(CARGO_RS_CHECK_TOOLCHAIN) run --profile $(CARGO_PROFILE) \
+	--example sha256_bool \
+	--features=$(TARGET_ARCH_FEATURE),boolean
+
 .PHONY: pcc # pcc stands for pre commit checks
-pcc: no_tfhe_typo check_fmt doc clippy_all check_compile_tests
+pcc: no_tfhe_typo no_dbg_log check_fmt doc clippy_all check_compile_tests
+
+.PHONY: fpcc # pcc stands for pre commit checks, the f stands for fast
+fpcc: no_tfhe_typo no_dbg_log check_fmt doc clippy_fast check_compile_tests
 
 .PHONY: conformance # Automatically fix problems that can be fixed
 conformance: fmt

README.md

@@ -1,31 +1,25 @@
 <p align="center">
 <!-- product name logo -->
-  <img width=600 src="https://user-images.githubusercontent.com/86411313/201107820-b1b861be-6b3f-46cc-bccd-ed051201781a.png">
+  <img width=600 src="https://user-images.githubusercontent.com/5758427/231206749-8f146b97-3c5a-4201-8388-3ffa88580415.png">
+</p>
+<hr/>
+<p align="center">
+  <a href="https://docs.zama.ai/tfhe-rs"> 📒 Read documentation</a> | <a href="https://zama.ai/community"> 💛 Community support</a>
 </p>
 <p align="center">
 <!-- Version badge using shields.io -->
   <a href="https://github.com/zama-ai/tfhe-rs/releases">
     <img src="https://img.shields.io/github/v/release/zama-ai/tfhe-rs?style=flat-square">
   </a>
-<!-- Link to docs badge using shields.io -->
-  <a href="https://docs.zama.ai/tfhe-rs">
-    <img src="https://img.shields.io/badge/read-documentation-yellow?style=flat-square">
-  </a>
-<!-- Community forum badge using shields.io -->
-  <a href="https://community.zama.ai">
-    <img src="https://img.shields.io/badge/community%20forum-online-brightgreen?style=flat-square">
-  </a>
-<!-- Open source badge using shields.io -->
-  <a href="https://docs.zama.ai/tfhe-rs/developers/contributing">
-    <img src="https://img.shields.io/badge/we're%20open%20source-contributing.md-blue?style=flat-square">
-  </a>
-<!-- Follow on twitter badge using shields.io -->
-  <a href="https://twitter.com/zama_fhe">
-    <img src="https://img.shields.io/badge/follow-zama_fhe-blue?logo=twitter&style=flat-square">
+<!-- Zama Bounty Program -->
+  <a href="https://github.com/zama-ai/bounty-program">
+    <img src="https://img.shields.io/badge/Contribute-Zama%20Bounty%20Program-yellow?style=flat-square">
   </a>
 </p>
+<hr/>
 
-**TFHE-rs** is a pure Rust implementation of TFHE for boolean and small integer
+
+**TFHE-rs** is a pure Rust implementation of TFHE for boolean and integer
 arithmetics over encrypted data. It includes:
  - a **Rust** API
  - a **C** API
@@ -37,19 +31,21 @@ implementation. The goal is to have a stable, simple, high-performance, and
 production-ready library for all the advanced features of TFHE.
 
 ## Getting Started
+The steps to run a first example are described below. 
 
+### Cargo.toml configuration
 To use the latest version of `TFHE-rs` in your project, you first need to add it as a dependency in your `Cargo.toml`:
 
 + For x86_64-based machines running Unix-like OSes:
 
 ```toml
-tfhe = { version = "*", features = ["boolean", "shortint", "x86_64-unix"] }
+tfhe = { version = "*", features = ["boolean", "shortint", "integer", "x86_64-unix"] }
 ```
 
 + For Apple Silicon or aarch64-based machines running Unix-like OSes:
 
 ```toml
-tfhe = { version = "*", features = ["boolean", "shortint", "aarch64-unix"] }
+tfhe = { version = "*", features = ["boolean", "shortint", "integer", "aarch64-unix"] }
 ```
 Note: users with ARM devices must use `TFHE-rs` by compiling using the `nightly` toolchain.
 
@@ -58,71 +54,74 @@ Note: users with ARM devices must use `TFHE-rs` by compiling using the `nightly`
 running Windows:
 
 ```toml
-tfhe = { version = "*", features = ["boolean", "shortint", "x86_64"] }
+tfhe = { version = "*", features = ["boolean", "shortint", "integer", "x86_64"] }
 ```
 
 Note: aarch64-based machines are not yet supported for Windows as it's currently missing an entropy source to be able to seed the [CSPRNGs](https://en.wikipedia.org/wiki/Cryptographically_secure_pseudorandom_number_generator) used in TFHE-rs
 
-Here is a full example evaluating a Boolean circuit:
 
-```rust
-use tfhe::boolean::prelude::*;
+## A simple example
 
-fn main() {
-// We generate a set of client/server keys, using the default parameters:
-    let (mut client_key, mut server_key) = gen_keys();
+Here is a full example:
 
-// We use the client secret key to encrypt two messages:
-    let ct_1 = client_key.encrypt(true);
-    let ct_2 = client_key.encrypt(false);
+``` rust
+use tfhe::prelude::*;
+use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint32, FheUint8};
 
-// We use the server public key to execute a boolean circuit:
-// if ((NOT ct_2) NAND (ct_1 AND ct_2)) then (NOT ct_2) else (ct_1 AND ct_2)
-    let ct_3 = server_key.not(&ct_2);
-    let ct_4 = server_key.and(&ct_1, &ct_2);
-    let ct_5 = server_key.nand(&ct_3, &ct_4);
-    let ct_6 = server_key.mux(&ct_5, &ct_3, &ct_4);
+fn main() -> Result<(), Box<dyn std::error::Error>> {
+    // Basic configuration to use homomorphic integers
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
 
-// We use the client key to decrypt the output of the circuit:
-    let output = client_key.decrypt(&ct_6);
-    assert_eq!(output, true);
+    // Key generation
+    let (client_key, server_keys) = generate_keys(config);
+
+    let clear_a = 1344u32;
+    let clear_b = 5u32;
+    let clear_c = 7u8;
+
+    // Encrypting the input data using the (private) client_key
+    // FheUint32: Encrypted equivalent to u32
+    let mut encrypted_a = FheUint32::try_encrypt(clear_a, &client_key)?;
+    let encrypted_b = FheUint32::try_encrypt(clear_b, &client_key)?;
+
+    // FheUint8: Encrypted equivalent to u8
+    let encrypted_c = FheUint8::try_encrypt(clear_c, &client_key)?;
+
+    // On the server side:
+    set_server_key(server_keys);
+
+    // Clear equivalent computations: 1344 * 8 = 10752
+    let encrypted_res_mul = &encrypted_a * &encrypted_b;
+
+    // Clear equivalent computations: 1344 >> 8 = 42
+    encrypted_a = &encrypted_res_mul >> &encrypted_b;
+
+    // Clear equivalent computations: let casted_a = a as u8;
+    let casted_a: FheUint8 = encrypted_a.cast_into();
+
+    // Clear equivalent computations: min(42, 7) = 7
+    let encrypted_res_min = &casted_a.min(&encrypted_c);
+
+    // Operation between clear and encrypted data:
+    // Clear equivalent computations: 7 & 1 = 1
+    let encrypted_res = encrypted_res_min & 1_u8;
+
+    // Decrypting on the client side:
+    let clear_res: u8 = encrypted_res.decrypt(&client_key);
+    assert_eq!(clear_res, 1_u8);
+
+    Ok(())
 }
 ```
 
-Another example of how the library can be used with shortints:
+To run this code, use the following command: 
+<p align="center"> <code> cargo run --release </code> </p>
 
-```rust
-use tfhe::shortint::prelude::*;
+Note that when running code that uses `tfhe-rs`, it is highly recommended
+to run in release mode with cargo's `--release` flag to have the best performances possible,
 
-fn main() {
-    // Generate a set of client/server keys, using the default parameters:
-    let (client_key, server_key) = gen_keys(Parameters::default());
-
-    let msg1 = 3;
-    let msg2 = 2;
-
-    // Encrypt two messages using the (private) client key:
-    let ct_1 = client_key.encrypt(msg1);
-    let ct_2 = client_key.encrypt(msg2);
-
-    // Homomorphically compute an addition
-    let ct_add = server_key.unchecked_add(&ct_1, &ct_2);
-
-    // Define the Hamming weight function
-    // f: x -> sum of the bits of x
-    let f = |x:u64| x.count_ones() as u64;
-
-    // Generate the accumulator for the function
-    let acc = server_key.generate_accumulator(f);
-
-    // Compute the function over the ciphertext using the PBS
-    let ct_res = server_key.keyswitch_programmable_bootstrap(&ct_add, &acc);
-
-    // Decrypt the ciphertext using the (private) client key
-    let output = client_key.decrypt(&ct_res);
-    assert_eq!(output, f(msg1 + msg2));
-}
-```
 
 ## Contributing
 
@@ -139,6 +138,24 @@ Only approved contributors can send pull requests, so please make sure to get in
 This library uses several dependencies and we would like to thank the contributors of those
 libraries.
 
+## Need support?
+<a target="_blank" href="https://community.zama.ai">
+  <img src="https://user-images.githubusercontent.com/5758427/231115030-21195b55-2629-4c01-9809-be5059243999.png">
+</a>
+
+## Citing TFHE-rs
+
+To cite TFHE-rs in academic papers, please use the following entry:
+
+```text
+@Misc{TFHE-rs,
+  title={{TFHE-rs: A Pure Rust Implementation of the TFHE Scheme for Boolean and Integer Arithmetics Over Encrypted Data}},
+  author={Zama},
+  year={2022},
+  note={\url{https://github.com/zama-ai/tfhe-rs}},
+}
+```
+
 ## License
 
 This software is distributed under the BSD-3-Clause-Clear license. If you have any questions,

apps/trivium/Cargo.toml

@@ -0,0 +1,24 @@
+[package]
+name = "tfhe-trivium"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+rayon = { version = "1.7.0"}
+
+[target.'cfg(target_arch = "x86_64")'.dependencies.tfhe]
+path = "../../tfhe"
+features = [ "boolean", "shortint", "integer", "x86_64" ]
+
+[target.'cfg(target_arch = "aarch64")'.dependencies.tfhe]
+path = "../../tfhe"
+features = [ "boolean", "shortint", "integer", "aarch64-unix" ]
+
+[dev-dependencies]
+criterion = { version = "0.4", features = [ "html_reports" ]}
+
+[[bench]]
+name = "trivium"
+harness = false

apps/trivium/README.md

@@ -0,0 +1,204 @@
+# FHE boolean Trivium implementation using TFHE-rs
+
+The cleartext boolean Trivium is available to be built using the function `TriviumStream::<bool>::new`. 
+This takes as input 2 arrays of 80 bool: the Trivium key and the IV. After initialization, it returns a TriviumStream on 
+which the user can call `next`, getting the next bit of the cipher stream, or `next_64`, which will compute 64 values at once,
+using multithreading to accelerate the computation.
+
+
+Quite similarly, the function `TriviumStream::<FheBool>::new` will return a very similar object running in FHE space. Its arguments are
+2 arrays of 80 FheBool representing the encrypted Trivium key, and the encrypted IV. It also requires a reference to the the server key of the 
+current scheme. This means that any user of this feature must also have the `tfhe-rs` crate as a dependency.
+
+
+Example of a Rust main below:
+```rust
+use tfhe::{ConfigBuilder, generate_keys, FheBool};
+use tfhe::prelude::*;
+
+use tfhe_trivium::TriviumStream;
+
+fn get_hexadecimal_string_from_lsb_first_stream(a: Vec<bool>) -> String {
+	assert!(a.len() % 8 == 0);
+	let mut hexadecimal: String = "".to_string();
+	for test in a.chunks(8) {
+		// Encoding is bytes in LSB order
+		match test[4..8] {
+			[false, false, false, false] => hexadecimal.push('0'),
+			[true, false, false, false] => hexadecimal.push('1'),
+			[false, true, false, false] => hexadecimal.push('2'),
+			[true, true, false, false] => hexadecimal.push('3'),
+
+			[false, false, true, false] => hexadecimal.push('4'),
+			[true, false, true, false] => hexadecimal.push('5'),
+			[false, true, true, false] => hexadecimal.push('6'),
+			[true, true, true, false] => hexadecimal.push('7'),
+
+			[false, false, false, true] => hexadecimal.push('8'),
+			[true, false, false, true] => hexadecimal.push('9'),
+			[false, true, false, true] => hexadecimal.push('A'),
+			[true, true, false, true] => hexadecimal.push('B'),
+
+			[false, false, true, true] => hexadecimal.push('C'),
+			[true, false, true, true] => hexadecimal.push('D'),
+			[false, true, true, true] => hexadecimal.push('E'),
+			[true, true, true, true] => hexadecimal.push('F'),
+			_ => ()
+		};
+		match test[0..4] {
+			[false, false, false, false] => hexadecimal.push('0'),
+			[true, false, false, false] => hexadecimal.push('1'),
+			[false, true, false, false] => hexadecimal.push('2'),
+			[true, true, false, false] => hexadecimal.push('3'),
+
+			[false, false, true, false] => hexadecimal.push('4'),
+			[true, false, true, false] => hexadecimal.push('5'),
+			[false, true, true, false] => hexadecimal.push('6'),
+			[true, true, true, false] => hexadecimal.push('7'),
+
+			[false, false, false, true] => hexadecimal.push('8'),
+			[true, false, false, true] => hexadecimal.push('9'),
+			[false, true, false, true] => hexadecimal.push('A'),
+			[true, true, false, true] => hexadecimal.push('B'),
+
+			[false, false, true, true] => hexadecimal.push('C'),
+			[true, false, true, true] => hexadecimal.push('D'),
+			[false, true, true, true] => hexadecimal.push('E'),
+			[true, true, true, true] => hexadecimal.push('F'),
+			_ => ()
+		};
+	}
+	return hexadecimal;
+}
+
+fn main() {
+	let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+	let (client_key, server_key) = generate_keys(config);
+
+	let key_string = "0053A6F94C9FF24598EB".to_string();
+	let mut key = [false; 80];
+
+	for i in (0..key_string.len()).step_by(2) {
+		let mut val: u8 = u8::from_str_radix(&key_string[i..i+2], 16).unwrap();
+		for j in 0..8 {
+			key[8*(i>>1) + j] = val % 2 == 1;
+			val >>= 1;
+		}
+	}
+
+	let iv_string = "0D74DB42A91077DE45AC".to_string();
+	let mut iv = [false; 80];
+
+	for i in (0..iv_string.len()).step_by(2) {
+		let mut val: u8 = u8::from_str_radix(&iv_string[i..i+2], 16).unwrap();
+		for j in 0..8 {
+			iv[8*(i>>1) + j] = val % 2 == 1;
+			val >>= 1;
+		}
+	}
+	
+	let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+	let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+	let cipher_iv = iv.map(|x| FheBool::encrypt(x, &client_key));
+
+
+	let mut trivium = TriviumStream::<FheBool>::new(cipher_key, cipher_iv, &server_key);
+
+	let mut vec = Vec::<bool>::with_capacity(64*8);
+	while vec.len() < 64*8 {
+		let cipher_outputs = trivium.next_64();
+		for c in cipher_outputs {
+			vec.push(c.decrypt(&client_key))
+		}
+	}
+
+	let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+	assert_eq!(output_0_63, hexadecimal[0..64*2]);
+}
+```
+
+# FHE byte Trivium implementation
+
+The same objects have also been implemented to stream bytes insead of booleans. They can be constructed and used in the same way via the functions `TriviumStreamByte::<u8>::new` and 
+`TriviumStreamByte::<FheUint8>::new` with the same arguments as before. The `FheUint8` version is significantly slower than the `FheBool` version, because not running 
+with the same cryptographic parameters. Its interest lie in its trans-ciphering capabilities: `TriviumStreamByte<FheUint8>` implements the trait `TransCiphering`, 
+meaning it implements the functions `trans_encrypt_64`. This function takes as input a `FheUint64` and outputs a `FheUint64`, the output being
+encrypted via tfhe and trivium. For convenience we also provide `trans_decrypt_64`, but this is of course the exact same function.
+
+Other sizes than 64 bit are expected to be available in the future.
+
+# FHE shortint Trivium implementation
+
+The same implementation is also available for generic Ciphertexts representing bits (meant to be used with parameters `PARAM_MESSAGE_1_CARRY_1_KS_PBS`). It uses a lower level API 
+of tfhe-rs, so the syntax is a little bit different. It also implements the `TransCiphering` trait. For optimization purposes, it does not internally run on the same 
+cryptographic parameters as the high level API of tfhe-rs. As such, it requires the usage of a casting key, to switch from one parameter space to another, which makes 
+its setup a little more intricate.
+
+Example code:
+```rust
+use tfhe::shortint::prelude::*;
+use tfhe::shortint::CastingKey;
+
+use tfhe::{ConfigBuilder, generate_keys, FheUint64};
+use tfhe::prelude::*;
+
+use tfhe_trivium::TriviumStreamShortint;
+
+fn test_shortint() {
+	let config = ConfigBuilder::all_disabled().enable_default_integers().build();
+	let (hl_client_key, hl_server_key) = generate_keys(config);
+	let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+	let ksk = CastingKey::new((&client_key, &server_key), (&hl_client_key, &hl_server_key));
+
+	let key_string = "0053A6F94C9FF24598EB".to_string();
+	let mut key = [0; 80];
+
+	for i in (0..key_string.len()).step_by(2) {
+		let mut val = u64::from_str_radix(&key_string[i..i+2], 16).unwrap();
+		for j in 0..8 {
+			key[8*(i>>1) + j] = val % 2;
+			val >>= 1;
+		}
+	}
+
+	let iv_string = "0D74DB42A91077DE45AC".to_string();
+	let mut iv = [0; 80];
+
+	for i in (0..iv_string.len()).step_by(2) {
+		let mut val = u64::from_str_radix(&iv_string[i..i+2], 16).unwrap();
+		for j in 0..8 {
+			iv[8*(i>>1) + j] = val % 2;
+			val >>= 1;
+		}
+	}
+	let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+	let cipher_key = key.map(|x| client_key.encrypt(x));
+	let cipher_iv = iv.map(|x| client_key.encrypt(x));
+
+	let mut ciphered_message = vec![FheUint64::try_encrypt(0u64, &hl_client_key).unwrap(); 9];
+
+	let mut trivium = TriviumStreamShortint::new(cipher_key, cipher_iv, &server_key, &ksk);
+
+	let mut vec = Vec::<u64>::with_capacity(8);
+	while vec.len() < 8 {
+		let trans_ciphered_message = trivium.trans_encrypt_64(ciphered_message.pop().unwrap(), &hl_server_key);
+		vec.push(trans_ciphered_message.decrypt(&hl_client_key));
+	}
+
+	let hexadecimal = get_hexagonal_string_from_u64(vec);
+	assert_eq!(output_0_63, hexadecimal[0..64*2]);
+}
+```
+
+# FHE Kreyvium implementation using tfhe-rs crate
+
+This will work in exactly the same way as the Trivium implementation, except that the key and iv need to be 128 bits now. Available for the same internal types as Trivium, with similar syntax.
+
+`KreyviumStreamByte<FheUint8>` and `KreyviumStreamShortint` also implement the `TransCiphering` trait.
+
+# Testing
+
+If you wish to run tests on this app, please run `cargo test -r trivium -- --test-threads=1` as multithreading provokes interferences between several running 
+Triviums at the same time.

apps/trivium/benches/kreyvium_bool.rs

@@ -0,0 +1,75 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheBool};
+
+use tfhe_trivium::KreyviumStream;
+
+use criterion::Criterion;
+
+pub fn kreyvium_bool_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [false; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [false; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+
+    let mut kreyvium = KreyviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("kreyvium bool generate 64 bits", |b| {
+        b.iter(|| kreyvium.next_64())
+    });
+}
+
+pub fn kreyvium_bool_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [false; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [false; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    c.bench_function("kreyvium bool warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+            let _kreyvium = KreyviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+        })
+    });
+}

apps/trivium/benches/kreyvium_byte.rs

@@ -0,0 +1,96 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheUint64, FheUint8};
+
+use tfhe_trivium::{KreyviumStreamByte, TransCiphering};
+
+use criterion::Criterion;
+
+pub fn kreyvium_byte_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .enable_function_evaluation_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let mut kreyvium = KreyviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("kreyvium byte generate 64 bits", |b| {
+        b.iter(|| kreyvium.next_64())
+    });
+}
+
+pub fn kreyvium_byte_trans(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .enable_function_evaluation_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &client_key).unwrap();
+    let mut kreyvium = KreyviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("kreyvium byte transencrypt 64 bits", |b| {
+        b.iter(|| kreyvium.trans_encrypt_64(ciphered_message.clone()))
+    });
+}
+
+pub fn kreyvium_byte_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .enable_function_evaluation_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    c.bench_function("kreyvium byte warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+            let _kreyvium = KreyviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+        })
+    });
+}

apps/trivium/benches/kreyvium_shortint.rs

@@ -0,0 +1,155 @@
+use tfhe::prelude::*;
+use tfhe::shortint::prelude::*;
+use tfhe::shortint::KeySwitchingKey;
+use tfhe::{generate_keys, ConfigBuilder, FheUint64};
+
+use tfhe_trivium::{KreyviumStreamShortint, TransCiphering};
+
+use criterion::Criterion;
+
+pub fn kreyvium_shortint_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    c.bench_function("kreyvium 1_1 warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| client_key.encrypt(x));
+            let _kreyvium = KreyviumStreamShortint::new(
+                cipher_key,
+                iv,
+                server_key.clone(),
+                ksk.clone(),
+                hl_server_key.clone(),
+            );
+        })
+    });
+}
+
+pub fn kreyvium_shortint_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let mut kreyvium = KreyviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    c.bench_function("kreyvium 1_1 generate 64 bits", |b| {
+        b.iter(|| kreyvium.next_64())
+    });
+}
+
+pub fn kreyvium_shortint_trans(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &hl_client_key).unwrap();
+    let mut kreyvium = KreyviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    c.bench_function("kreyvium 1_1 transencrypt 64 bits", |b| {
+        b.iter(|| kreyvium.trans_encrypt_64(ciphered_message.clone()))
+    });
+}

apps/trivium/benches/trivium.rs

@@ -0,0 +1,53 @@
+use criterion::{criterion_group, criterion_main};
+
+mod trivium_bool;
+criterion_group!(
+    trivium_bool,
+    trivium_bool::trivium_bool_gen,
+    trivium_bool::trivium_bool_warmup
+);
+mod kreyvium_bool;
+criterion_group!(
+    kreyvium_bool,
+    kreyvium_bool::kreyvium_bool_gen,
+    kreyvium_bool::kreyvium_bool_warmup
+);
+
+mod trivium_shortint;
+criterion_group!(
+    trivium_shortint,
+    trivium_shortint::trivium_shortint_gen,
+    trivium_shortint::trivium_shortint_warmup,
+    trivium_shortint::trivium_shortint_trans
+);
+mod kreyvium_shortint;
+criterion_group!(
+    kreyvium_shortint,
+    kreyvium_shortint::kreyvium_shortint_gen,
+    kreyvium_shortint::kreyvium_shortint_warmup,
+    kreyvium_shortint::kreyvium_shortint_trans
+);
+
+mod trivium_byte;
+criterion_group!(
+    trivium_byte,
+    trivium_byte::trivium_byte_gen,
+    trivium_byte::trivium_byte_trans,
+    trivium_byte::trivium_byte_warmup
+);
+mod kreyvium_byte;
+criterion_group!(
+    kreyvium_byte,
+    kreyvium_byte::kreyvium_byte_gen,
+    kreyvium_byte::kreyvium_byte_trans,
+    kreyvium_byte::kreyvium_byte_warmup
+);
+
+criterion_main!(
+    trivium_bool,
+    trivium_shortint,
+    trivium_byte,
+    kreyvium_bool,
+    kreyvium_shortint,
+    kreyvium_byte,
+);

apps/trivium/benches/trivium_bool.rs

@@ -0,0 +1,75 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheBool};
+
+use tfhe_trivium::TriviumStream;
+
+use criterion::Criterion;
+
+pub fn trivium_bool_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [false; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [false; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+
+    let mut trivium = TriviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("trivium bool generate 64 bits", |b| {
+        b.iter(|| trivium.next_64())
+    });
+}
+
+pub fn trivium_bool_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [false; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [false; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    c.bench_function("trivium bool warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+            let _trivium = TriviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+        })
+    });
+}

apps/trivium/benches/trivium_byte.rs

@@ -0,0 +1,93 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheUint64, FheUint8};
+
+use tfhe_trivium::{TransCiphering, TriviumStreamByte};
+
+use criterion::Criterion;
+
+pub fn trivium_byte_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let mut trivium = TriviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("trivium byte generate 64 bits", |b| {
+        b.iter(|| trivium.next_64())
+    });
+}
+
+pub fn trivium_byte_trans(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &client_key).unwrap();
+    let mut trivium = TriviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    c.bench_function("trivium byte transencrypt 64 bits", |b| {
+        b.iter(|| trivium.trans_encrypt_64(ciphered_message.clone()))
+    });
+}
+
+pub fn trivium_byte_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    c.bench_function("trivium byte warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+            let _trivium = TriviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+        })
+    });
+}

apps/trivium/benches/trivium_shortint.rs

@@ -0,0 +1,155 @@
+use tfhe::prelude::*;
+use tfhe::shortint::prelude::*;
+use tfhe::shortint::KeySwitchingKey;
+use tfhe::{generate_keys, ConfigBuilder, FheUint64};
+
+use tfhe_trivium::{TransCiphering, TriviumStreamShortint};
+
+use criterion::Criterion;
+
+pub fn trivium_shortint_warmup(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    c.bench_function("trivium 1_1 warmup", |b| {
+        b.iter(|| {
+            let cipher_key = key.map(|x| client_key.encrypt(x));
+            let _trivium = TriviumStreamShortint::new(
+                cipher_key,
+                iv,
+                server_key.clone(),
+                ksk.clone(),
+                hl_server_key.clone(),
+            );
+        })
+    });
+}
+
+pub fn trivium_shortint_gen(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let mut trivium = TriviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    c.bench_function("trivium 1_1 generate 64 bits", |b| {
+        b.iter(|| trivium.next_64())
+    });
+}
+
+pub fn trivium_shortint_trans(c: &mut Criterion) {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &hl_client_key).unwrap();
+    let mut trivium = TriviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    c.bench_function("trivium 1_1 transencrypt 64 bits", |b| {
+        b.iter(|| trivium.trans_encrypt_64(ciphered_message.clone()))
+    });
+}

apps/trivium/src/kreyvium/kreyvium.rs

@@ -0,0 +1,257 @@
+//! This module implements the Kreyvium stream cipher, using booleans or FheBool
+//! for the representaion of the inner bits.
+
+use crate::static_deque::StaticDeque;
+
+use tfhe::prelude::*;
+use tfhe::{set_server_key, unset_server_key, FheBool, ServerKey};
+
+use rayon::prelude::*;
+
+/// Internal trait specifying which operations are necessary for KreyviumStream generic type
+pub trait KreyviumBoolInput<OpOutput>:
+    Sized
+    + Clone
+    + std::ops::BitXor<Output = OpOutput>
+    + std::ops::BitAnd<Output = OpOutput>
+    + std::ops::Not<Output = OpOutput>
+{
+}
+impl KreyviumBoolInput<bool> for bool {}
+impl KreyviumBoolInput<bool> for &bool {}
+impl KreyviumBoolInput<FheBool> for FheBool {}
+impl KreyviumBoolInput<FheBool> for &FheBool {}
+
+/// KreyviumStream: a struct implementing the Kreyvium stream cipher, using T for the internal
+/// representation of bits (bool or FheBool). To be able to compute FHE operations, it also owns
+/// an Option for a ServerKey.
+pub struct KreyviumStream<T> {
+    a: StaticDeque<93, T>,
+    b: StaticDeque<84, T>,
+    c: StaticDeque<111, T>,
+    k: StaticDeque<128, T>,
+    iv: StaticDeque<128, T>,
+    fhe_key: Option<ServerKey>,
+}
+
+impl KreyviumStream<bool> {
+    /// Contructor for `KreyviumStream<bool>`: arguments are the secret key and the input vector.
+    /// Outputs a KreyviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(mut key: [bool; 128], mut iv: [bool; 128]) -> KreyviumStream<bool> {
+        // Initialization of Kreyvium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register = [false; 93];
+        let mut b_register = [false; 84];
+        let mut c_register = [false; 111];
+
+        for i in 0..93 {
+            a_register[i] = key[128 - 93 + i];
+        }
+        for i in 0..84 {
+            b_register[i] = iv[128 - 84 + i];
+        }
+        for i in 0..44 {
+            c_register[111 - 44 + i] = iv[i];
+        }
+        for i in 0..66 {
+            c_register[i + 1] = true;
+        }
+
+        key.reverse();
+        iv.reverse();
+        KreyviumStream::<bool>::new_from_registers(
+            a_register, b_register, c_register, key, iv, None,
+        )
+    }
+}
+
+impl KreyviumStream<FheBool> {
+    /// Constructor for `KreyviumStream<FheBool>`: arguments are the encrypted secret key and input
+    /// vector, and the FHE server key.
+    /// Outputs a KreyviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(
+        mut key: [FheBool; 128],
+        mut iv: [bool; 128],
+        sk: &ServerKey,
+    ) -> KreyviumStream<FheBool> {
+        set_server_key(sk.clone());
+
+        // Initialization of Kreyvium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register = [false; 93].map(|x| FheBool::encrypt_trivial(x));
+        let mut b_register = [false; 84].map(|x| FheBool::encrypt_trivial(x));
+        let mut c_register = [false; 111].map(|x| FheBool::encrypt_trivial(x));
+
+        for i in 0..93 {
+            a_register[i] = key[128 - 93 + i].clone();
+        }
+        for i in 0..84 {
+            b_register[i] = FheBool::encrypt_trivial(iv[128 - 84 + i]);
+        }
+        for i in 0..44 {
+            c_register[111 - 44 + i] = FheBool::encrypt_trivial(iv[i]);
+        }
+        for i in 0..66 {
+            c_register[i + 1] = FheBool::encrypt_trivial(true);
+        }
+
+        key.reverse();
+        iv.reverse();
+        let iv = iv.map(|x| FheBool::encrypt_trivial(x));
+
+        unset_server_key();
+        KreyviumStream::<FheBool>::new_from_registers(
+            a_register,
+            b_register,
+            c_register,
+            key,
+            iv,
+            Some(sk.clone()),
+        )
+    }
+}
+
+impl<T> KreyviumStream<T>
+where
+    T: KreyviumBoolInput<T> + std::marker::Send + std::marker::Sync,
+    for<'a> &'a T: KreyviumBoolInput<T>,
+{
+    /// Internal generic contructor: arguments are already prepared registers, and an optional FHE
+    /// server key
+    fn new_from_registers(
+        a_register: [T; 93],
+        b_register: [T; 84],
+        c_register: [T; 111],
+        k_register: [T; 128],
+        iv_register: [T; 128],
+        key: Option<ServerKey>,
+    ) -> Self {
+        let mut ret = Self {
+            a: StaticDeque::<93, T>::new(a_register),
+            b: StaticDeque::<84, T>::new(b_register),
+            c: StaticDeque::<111, T>::new(c_register),
+            k: StaticDeque::<128, T>::new(k_register),
+            iv: StaticDeque::<128, T>::new(iv_register),
+            fhe_key: key,
+        };
+        ret.init();
+        ret
+    }
+
+    /// The specification of Kreyvium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes one turn of the stream, updating registers and outputting the new bit.
+    pub fn next(&mut self) -> T {
+        match &self.fhe_key {
+            Some(sk) => set_server_key(sk.clone()),
+            None => (),
+        };
+
+        let [o, a, b, c] = self.get_output_and_values(0);
+
+        self.a.push(a);
+        self.b.push(b);
+        self.c.push(c);
+        self.k.shift();
+        self.iv.shift();
+
+        o
+    }
+
+    /// Computes a potential future step of Kreyvium, n terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, n: usize) -> [T; 4] {
+        assert!(n < 65);
+
+        let (((temp_a, temp_b), (temp_c, a_and)), (b_and, c_and)) = rayon::join(
+            || {
+                rayon::join(
+                    || {
+                        rayon::join(
+                            || &self.a[65 - n] ^ &self.a[92 - n],
+                            || &self.b[68 - n] ^ &self.b[83 - n],
+                        )
+                    },
+                    || {
+                        rayon::join(
+                            || &(&self.c[65 - n] ^ &self.c[110 - n]) ^ &self.k[127 - n],
+                            || &(&self.a[91 - n] & &self.a[90 - n]) ^ &self.iv[127 - n],
+                        )
+                    },
+                )
+            },
+            || {
+                rayon::join(
+                    || &self.b[82 - n] & &self.b[81 - n],
+                    || &self.c[109 - n] & &self.c[108 - n],
+                )
+            },
+        );
+
+        let ((o, a), (b, c)) = rayon::join(
+            || {
+                rayon::join(
+                    || &(&temp_a ^ &temp_b) ^ &temp_c,
+                    || &temp_c ^ &(&c_and ^ &self.a[68 - n]),
+                )
+            },
+            || {
+                rayon::join(
+                    || &temp_a ^ &(&a_and ^ &self.b[77 - n]),
+                    || &temp_b ^ &(&b_and ^ &self.c[86 - n]),
+                )
+            },
+        );
+
+        [o, a, b, c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 64 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[T; 4]> {
+        (0..64)
+            .into_par_iter()
+            .map(|x| self.get_output_and_values(x))
+            .rev()
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<T> {
+        match &self.fhe_key {
+            Some(sk) => {
+                rayon::broadcast(|_| set_server_key(sk.clone()));
+            }
+            None => (),
+        }
+        let mut values = self.get_64_output_and_values();
+        match &self.fhe_key {
+            Some(_) => {
+                rayon::broadcast(|_| unset_server_key());
+            }
+            None => (),
+        }
+
+        let mut ret = Vec::<T>::with_capacity(64);
+
+        while let Some([o, a, b, c]) = values.pop() {
+            ret.push(o);
+            self.a.push(a);
+            self.b.push(b);
+            self.c.push(c);
+        }
+        self.k.n_shifts(64);
+        self.iv.n_shifts(64);
+        ret
+    }
+}

apps/trivium/src/kreyvium/kreyvium_byte.rs

@@ -0,0 +1,297 @@
+//! This module implements the Kreyvium stream cipher, using u8 or FheUint8
+//! for the representaion of the inner bits.
+
+use crate::static_deque::{StaticByteDeque, StaticByteDequeInput};
+
+use tfhe::prelude::*;
+use tfhe::{set_server_key, unset_server_key, FheUint8, ServerKey};
+
+use rayon::prelude::*;
+
+/// Internal trait specifying which operations are necessary for KreyviumStreamByte generic type
+pub trait KreyviumByteInput<OpOutput>:
+    Sized
+    + Send
+    + Sync
+    + Clone
+    + StaticByteDequeInput<OpOutput>
+    + std::ops::BitXor<Output = OpOutput>
+    + std::ops::BitAnd<Output = OpOutput>
+    + std::ops::Shr<u8, Output = OpOutput>
+    + std::ops::Shl<u8, Output = OpOutput>
+    + std::ops::Add<Output = OpOutput>
+{
+}
+impl KreyviumByteInput<u8> for u8 {}
+impl KreyviumByteInput<u8> for &u8 {}
+impl KreyviumByteInput<FheUint8> for FheUint8 {}
+impl KreyviumByteInput<FheUint8> for &FheUint8 {}
+
+/// KreyviumStreamByte: a struct implementing the Kreyvium stream cipher, using T for the internal
+/// representation of bits (u8 or FheUint8). To be able to compute FHE operations, it also owns
+/// an Option for a ServerKey.
+/// Since the original Kreyvium registers' sizes are not a multiple of 8, these registers (which
+/// store byte-like objects) have a size that is the eigth of the closest multiple of 8 above the
+/// originals' sizes.
+pub struct KreyviumStreamByte<T> {
+    a_byte: StaticByteDeque<12, T>,
+    b_byte: StaticByteDeque<11, T>,
+    c_byte: StaticByteDeque<14, T>,
+    k_byte: StaticByteDeque<16, T>,
+    iv_byte: StaticByteDeque<16, T>,
+    fhe_key: Option<ServerKey>,
+}
+
+impl KreyviumStreamByte<u8> {
+    /// Contructor for `KreyviumStreamByte<u8>`: arguments are the secret key and the input vector.
+    /// Outputs a KreyviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(key_bytes: [u8; 16], iv_bytes: [u8; 16]) -> KreyviumStreamByte<u8> {
+        // Initialization of Kreyvium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_byte_reg = [0u8; 12];
+        let mut b_byte_reg = [0u8; 11];
+        let mut c_byte_reg = [0u8; 14];
+
+        // Copy key bits into a register
+        for b in 0..12 {
+            a_byte_reg[b] = key_bytes[b + 4];
+        }
+        // Copy iv bits into a register
+        for b in 0..11 {
+            b_byte_reg[b] = iv_bytes[b + 5];
+        }
+        // Copy a lot of ones in the c register
+        c_byte_reg[0] = 252;
+        for b in 1..8 {
+            c_byte_reg[b] = 255;
+        }
+        // Copy iv bits in the c register
+        c_byte_reg[8] = (iv_bytes[0] << 4) | 31;
+        for b in 9..14 {
+            c_byte_reg[b] = (iv_bytes[b - 9] >> 4) | (iv_bytes[b - 8] << 4);
+        }
+
+        // Key and iv are stored in reverse in their shift registers
+        let mut key = key_bytes.map(|b| b.reverse_bits());
+        let mut iv = iv_bytes.map(|b| b.reverse_bits());
+        key.reverse();
+        iv.reverse();
+
+        let mut ret = KreyviumStreamByte::<u8>::new_from_registers(
+            a_byte_reg, b_byte_reg, c_byte_reg, key, iv, None,
+        );
+        ret.init();
+        ret
+    }
+}
+
+impl KreyviumStreamByte<FheUint8> {
+    /// Constructor for `KreyviumStream<FheUint8>`: arguments are the encrypted secret key and input
+    /// vector, and the FHE server key.
+    /// Outputs a KreyviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(
+        key_bytes: [FheUint8; 16],
+        iv_bytes: [u8; 16],
+        server_key: &ServerKey,
+    ) -> KreyviumStreamByte<FheUint8> {
+        set_server_key(server_key.clone());
+
+        // Initialization of Kreyvium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_byte_reg = [0u8; 12].map(|x| FheUint8::encrypt_trivial(x));
+        let mut b_byte_reg = [0u8; 11].map(|x| FheUint8::encrypt_trivial(x));
+        let mut c_byte_reg = [0u8; 14].map(|x| FheUint8::encrypt_trivial(x));
+
+        // Copy key bits into a register
+        for b in 0..12 {
+            a_byte_reg[b] = key_bytes[b + 4].clone();
+        }
+        // Copy iv bits into a register
+        for b in 0..11 {
+            b_byte_reg[b] = FheUint8::encrypt_trivial(iv_bytes[b + 5]);
+        }
+        // Copy a lot of ones in the c register
+        c_byte_reg[0] = FheUint8::encrypt_trivial(252u8);
+        for b in 1..8 {
+            c_byte_reg[b] = FheUint8::encrypt_trivial(255u8);
+        }
+        // Copy iv bits in the c register
+        c_byte_reg[8] = FheUint8::encrypt_trivial((&iv_bytes[0] << 4u8) | 31u8);
+        for b in 9..14 {
+            c_byte_reg[b] =
+                FheUint8::encrypt_trivial((&iv_bytes[b - 9] >> 4u8) | (&iv_bytes[b - 8] << 4u8));
+        }
+
+        // Key and iv are stored in reverse in their shift registers
+        let mut key = key_bytes.map(|b| b.map(|x| (x as u8).reverse_bits() as u64));
+        let mut iv = iv_bytes.map(|x| FheUint8::encrypt_trivial(x.reverse_bits()));
+        key.reverse();
+        iv.reverse();
+
+        unset_server_key();
+
+        let mut ret = KreyviumStreamByte::<FheUint8>::new_from_registers(
+            a_byte_reg,
+            b_byte_reg,
+            c_byte_reg,
+            key,
+            iv,
+            Some(server_key.clone()),
+        );
+        ret.init();
+        ret
+    }
+}
+
+impl<T> KreyviumStreamByte<T>
+where
+    T: KreyviumByteInput<T> + Send,
+    for<'a> &'a T: KreyviumByteInput<T>,
+{
+    /// Internal generic contructor: arguments are already prepared registers, and an optional FHE
+    /// server key
+    fn new_from_registers(
+        a_register: [T; 12],
+        b_register: [T; 11],
+        c_register: [T; 14],
+        k_register: [T; 16],
+        iv_register: [T; 16],
+        sk: Option<ServerKey>,
+    ) -> Self {
+        Self {
+            a_byte: StaticByteDeque::<12, T>::new(a_register),
+            b_byte: StaticByteDeque::<11, T>::new(b_register),
+            c_byte: StaticByteDeque::<14, T>::new(c_register),
+            k_byte: StaticByteDeque::<16, T>::new(k_register),
+            iv_byte: StaticByteDeque::<16, T>::new(iv_register),
+            fhe_key: sk,
+        }
+    }
+
+    /// The specification of Kreyvium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes 8 potential future step of Kreyvium, b*8 terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, b: usize) -> [T; 4] {
+        let n = b * 8 + 7;
+        assert!(n < 65);
+
+        let (((k, iv), (a1, a2, a3, a4, a5)), ((b1, b2, b3, b4, b5), (c1, c2, c3, c4, c5))) =
+            rayon::join(
+                || {
+                    rayon::join(
+                        || (self.k_byte.byte(127 - n), self.iv_byte.byte(127 - n)),
+                        || Self::get_bytes(&self.a_byte, [91 - n, 90 - n, 68 - n, 65 - n, 92 - n]),
+                    )
+                },
+                || {
+                    rayon::join(
+                        || Self::get_bytes(&self.b_byte, [82 - n, 81 - n, 77 - n, 68 - n, 83 - n]),
+                        || {
+                            Self::get_bytes(
+                                &self.c_byte,
+                                [109 - n, 108 - n, 86 - n, 65 - n, 110 - n],
+                            )
+                        },
+                    )
+                },
+            );
+
+        let (((temp_a, temp_b), (temp_c, a_and)), (b_and, c_and)) = rayon::join(
+            || {
+                rayon::join(
+                    || rayon::join(|| a4 ^ a5, || b4 ^ b5),
+                    || rayon::join(|| c4 ^ c5 ^ k, || a1 & a2 ^ iv),
+                )
+            },
+            || rayon::join(|| b1 & b2, || c1 & c2),
+        );
+
+        let (temp_a_2, temp_b_2, temp_c_2) = (temp_a.clone(), temp_b.clone(), temp_c.clone());
+
+        let ((o, a), (b, c)) = rayon::join(
+            || {
+                rayon::join(
+                    || (temp_a_2 ^ temp_b_2) ^ temp_c_2,
+                    || temp_c ^ ((c_and) ^ a3),
+                )
+            },
+            || rayon::join(|| temp_a ^ (a_and ^ b3), || temp_b ^ (b_and ^ c3)),
+        );
+
+        [o, a, b, c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 8 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[T; 4]> {
+        (0..8)
+            .into_par_iter()
+            .map(|i| self.get_output_and_values(i))
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits (in 8 bytes) all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<T> {
+        match &self.fhe_key {
+            Some(sk) => {
+                rayon::broadcast(|_| set_server_key(sk.clone()));
+            }
+            None => (),
+        }
+        let values = self.get_64_output_and_values();
+        match &self.fhe_key {
+            Some(_) => {
+                rayon::broadcast(|_| unset_server_key());
+            }
+            None => (),
+        }
+
+        let mut bytes = Vec::<T>::with_capacity(8);
+        for [o, a, b, c] in values {
+            self.a_byte.push(a);
+            self.b_byte.push(b);
+            self.c_byte.push(c);
+            bytes.push(o);
+        }
+        self.k_byte.n_shifts(8);
+        self.iv_byte.n_shifts(8);
+
+        bytes
+    }
+
+    /// Reconstructs a bunch of 5 bytes in a parallel fashion.
+    fn get_bytes<const N: usize>(
+        reg: &StaticByteDeque<N, T>,
+        offsets: [usize; 5],
+    ) -> (T, T, T, T, T) {
+        let mut ret = offsets
+            .par_iter()
+            .rev()
+            .map(|&i| reg.byte(i))
+            .collect::<Vec<_>>();
+        (
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+        )
+    }
+}
+
+impl KreyviumStreamByte<FheUint8> {
+    pub fn get_server_key(&self) -> &ServerKey {
+        &self.fhe_key.as_ref().unwrap()
+    }
+}

apps/trivium/src/kreyvium/kreyvium_shortint.rs

@@ -0,0 +1,205 @@
+use crate::static_deque::StaticDeque;
+
+use tfhe::shortint::prelude::*;
+
+use rayon::prelude::*;
+
+/// KreyviumStreamShortint: a struct implementing the Kreyvium stream cipher, using a generic
+/// Ciphertext for the internal representation of bits (intended to represent a single bit). To be
+/// able to compute FHE operations, it also owns a ServerKey.
+pub struct KreyviumStreamShortint {
+    a: StaticDeque<93, Ciphertext>,
+    b: StaticDeque<84, Ciphertext>,
+    c: StaticDeque<111, Ciphertext>,
+    k: StaticDeque<128, Ciphertext>,
+    iv: StaticDeque<128, Ciphertext>,
+    internal_server_key: ServerKey,
+    transciphering_casting_key: KeySwitchingKey,
+    hl_server_key: tfhe::ServerKey,
+}
+
+impl KreyviumStreamShortint {
+    /// Contructor for KreyviumStreamShortint: arguments are the secret key and the input vector,
+    /// and a ServerKey reference. Outputs a KreyviumStream object already initialized (1152
+    /// steps have been run before returning)
+    pub fn new(
+        mut key: [Ciphertext; 128],
+        mut iv: [u64; 128],
+        sk: ServerKey,
+        ksk: KeySwitchingKey,
+        hl_sk: tfhe::ServerKey,
+    ) -> Self {
+        // Initialization of Kreyvium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register: [Ciphertext; 93] = [0; 93].map(|x| sk.create_trivial(x));
+        let mut b_register: [Ciphertext; 84] = [0; 84].map(|x| sk.create_trivial(x));
+        let mut c_register: [Ciphertext; 111] = [0; 111].map(|x| sk.create_trivial(x));
+
+        for i in 0..93 {
+            a_register[i] = key[128 - 93 + i].clone();
+        }
+        for i in 0..84 {
+            b_register[i] = sk.create_trivial(iv[128 - 84 + i]);
+        }
+        for i in 0..44 {
+            c_register[111 - 44 + i] = sk.create_trivial(iv[i]);
+        }
+        for i in 0..66 {
+            c_register[i + 1] = sk.create_trivial(1);
+        }
+
+        key.reverse();
+        iv.reverse();
+        let iv = iv.map(|x| sk.create_trivial(x));
+
+        let mut ret = Self {
+            a: StaticDeque::<93, Ciphertext>::new(a_register),
+            b: StaticDeque::<84, Ciphertext>::new(b_register),
+            c: StaticDeque::<111, Ciphertext>::new(c_register),
+            k: StaticDeque::<128, Ciphertext>::new(key),
+            iv: StaticDeque::<128, Ciphertext>::new(iv),
+            internal_server_key: sk,
+            transciphering_casting_key: ksk,
+            hl_server_key: hl_sk,
+        };
+        ret.init();
+        ret
+    }
+
+    /// The specification of Kreyvium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes one turn of the stream, updating registers and outputting the new bit.
+    pub fn next(&mut self) -> Ciphertext {
+        let [o, a, b, c] = self.get_output_and_values(0);
+
+        self.a.push(a);
+        self.b.push(b);
+        self.c.push(c);
+
+        o
+    }
+
+    /// Computes a potential future step of Kreyvium, n terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, n: usize) -> [Ciphertext; 4] {
+        let (k, iv) = (&self.k[127 - n], &self.iv[127 - n]);
+
+        let (a1, a2, a3, a4, a5) = (
+            &self.a[65 - n],
+            &self.a[92 - n],
+            &self.a[91 - n],
+            &self.a[90 - n],
+            &self.a[68 - n],
+        );
+        let (b1, b2, b3, b4, b5) = (
+            &self.b[68 - n],
+            &self.b[83 - n],
+            &self.b[82 - n],
+            &self.b[81 - n],
+            &self.b[77 - n],
+        );
+        let (c1, c2, c3, c4, c5) = (
+            &self.c[65 - n],
+            &self.c[110 - n],
+            &self.c[109 - n],
+            &self.c[108 - n],
+            &self.c[86 - n],
+        );
+
+        let temp_a = self.internal_server_key.unchecked_add(a1, a2);
+        let temp_b = self.internal_server_key.unchecked_add(b1, b2);
+        let mut temp_c = self.internal_server_key.unchecked_add(c1, c2);
+        self.internal_server_key
+            .unchecked_add_assign(&mut temp_c, k);
+
+        let ((new_a, new_b), (new_c, o)) = rayon::join(
+            || {
+                rayon::join(
+                    || {
+                        let mut new_a = self.internal_server_key.unchecked_bitand(c3, c4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_a, a5);
+                        self.internal_server_key.add_assign(&mut new_a, &temp_c);
+                        new_a
+                    },
+                    || {
+                        let mut new_b = self.internal_server_key.unchecked_bitand(a3, a4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_b, b5);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_b, &temp_a);
+                        self.internal_server_key.add_assign(&mut new_b, iv);
+                        new_b
+                    },
+                )
+            },
+            || {
+                rayon::join(
+                    || {
+                        let mut new_c = self.internal_server_key.unchecked_bitand(b3, b4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_c, c5);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_c, &temp_b);
+                        self.internal_server_key.clear_carry_assign(&mut new_c);
+                        new_c
+                    },
+                    || {
+                        self.internal_server_key.bitxor(
+                            &self.internal_server_key.unchecked_add(&temp_a, &temp_b),
+                            &temp_c,
+                        )
+                    },
+                )
+            },
+        );
+
+        [o, new_a, new_b, new_c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 64 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[Ciphertext; 4]> {
+        (0..64)
+            .into_par_iter()
+            .map(|x| self.get_output_and_values(x))
+            .rev()
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<Ciphertext> {
+        let mut values = self.get_64_output_and_values();
+
+        let mut ret = Vec::<Ciphertext>::with_capacity(64);
+        while let Some([o, a, b, c]) = values.pop() {
+            ret.push(o);
+            self.a.push(a);
+            self.b.push(b);
+            self.c.push(c);
+        }
+        self.k.n_shifts(64);
+        self.iv.n_shifts(64);
+        ret
+    }
+
+    pub fn get_internal_server_key(&self) -> &ServerKey {
+        &self.internal_server_key
+    }
+
+    pub fn get_casting_key(&self) -> &KeySwitchingKey {
+        &self.transciphering_casting_key
+    }
+
+    pub fn get_hl_server_key(&self) -> &tfhe::ServerKey {
+        &self.hl_server_key
+    }
+}

apps/trivium/src/kreyvium/mod.rs

@@ -0,0 +1,11 @@
+mod kreyvium;
+pub use kreyvium::KreyviumStream;
+
+mod kreyvium_byte;
+pub use kreyvium_byte::KreyviumStreamByte;
+
+mod kreyvium_shortint;
+pub use kreyvium_shortint::KreyviumStreamShortint;
+
+#[cfg(test)]
+mod test;

apps/trivium/src/kreyvium/test.rs

@@ -0,0 +1,378 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheBool, FheUint64, FheUint8};
+
+use crate::{KreyviumStream, KreyviumStreamByte, KreyviumStreamShortint, TransCiphering};
+
+// Values for these tests come from the github repo renaud1239/Kreyvium,
+// commit fd6828f68711276c25f55e605935028f5e843f43
+
+fn get_hexadecimal_string_from_lsb_first_stream(a: Vec<bool>) -> String {
+    assert!(a.len() % 8 == 0);
+    let mut hexadecimal: String = "".to_string();
+    for test in a.chunks(8) {
+        // Encoding is bytes in LSB order
+        match test[4..8] {
+            [false, false, false, false] => hexadecimal.push('0'),
+            [true, false, false, false] => hexadecimal.push('1'),
+            [false, true, false, false] => hexadecimal.push('2'),
+            [true, true, false, false] => hexadecimal.push('3'),
+
+            [false, false, true, false] => hexadecimal.push('4'),
+            [true, false, true, false] => hexadecimal.push('5'),
+            [false, true, true, false] => hexadecimal.push('6'),
+            [true, true, true, false] => hexadecimal.push('7'),
+
+            [false, false, false, true] => hexadecimal.push('8'),
+            [true, false, false, true] => hexadecimal.push('9'),
+            [false, true, false, true] => hexadecimal.push('A'),
+            [true, true, false, true] => hexadecimal.push('B'),
+
+            [false, false, true, true] => hexadecimal.push('C'),
+            [true, false, true, true] => hexadecimal.push('D'),
+            [false, true, true, true] => hexadecimal.push('E'),
+            [true, true, true, true] => hexadecimal.push('F'),
+            _ => (),
+        };
+        match test[0..4] {
+            [false, false, false, false] => hexadecimal.push('0'),
+            [true, false, false, false] => hexadecimal.push('1'),
+            [false, true, false, false] => hexadecimal.push('2'),
+            [true, true, false, false] => hexadecimal.push('3'),
+
+            [false, false, true, false] => hexadecimal.push('4'),
+            [true, false, true, false] => hexadecimal.push('5'),
+            [false, true, true, false] => hexadecimal.push('6'),
+            [true, true, true, false] => hexadecimal.push('7'),
+
+            [false, false, false, true] => hexadecimal.push('8'),
+            [true, false, false, true] => hexadecimal.push('9'),
+            [false, true, false, true] => hexadecimal.push('A'),
+            [true, true, false, true] => hexadecimal.push('B'),
+
+            [false, false, true, true] => hexadecimal.push('C'),
+            [true, false, true, true] => hexadecimal.push('D'),
+            [false, true, true, true] => hexadecimal.push('E'),
+            [true, true, true, true] => hexadecimal.push('F'),
+            _ => (),
+        };
+    }
+    return hexadecimal;
+}
+
+fn get_hexagonal_string_from_bytes(a: Vec<u8>) -> String {
+    assert!(a.len() % 8 == 0);
+    let mut hexadecimal: String = "".to_string();
+    for test in a {
+        hexadecimal.push_str(&format!("{:02X?}", test));
+    }
+    return hexadecimal;
+}
+
+fn get_hexagonal_string_from_u64(a: Vec<u64>) -> String {
+    let mut hexadecimal: String = "".to_string();
+    for test in a {
+        hexadecimal.push_str(&format!("{:016X?}", test));
+    }
+    return hexadecimal;
+}
+
+#[test]
+fn kreyvium_test_1() {
+    let key = [false; 128];
+    let iv = [false; 128];
+    let output = "26DCF1F4BC0F1922";
+
+    let mut kreyvium = KreyviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(64);
+    while vec.len() < 64 {
+        vec.push(kreyvium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_2() {
+    let mut key = [false; 128];
+    let iv = [false; 128];
+    key[0] = true;
+
+    let output = "4FD421D4DA3D2C8A";
+
+    let mut kreyvium = KreyviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(64);
+    while vec.len() < 64 {
+        vec.push(kreyvium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_3() {
+    let key = [false; 128];
+    let mut iv = [false; 128];
+    iv[0] = true;
+
+    let output = "C9217BA0D762ACA1";
+
+    let mut kreyvium = KreyviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(64);
+    while vec.len() < 64 {
+        vec.push(kreyvium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_4() {
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [false; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [false; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let output = "D1F0303482061111";
+
+    let mut kreyvium = KreyviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(64);
+    while vec.len() < 64 {
+        vec.push(kreyvium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(hexadecimal, output);
+}
+
+#[test]
+fn kreyvium_test_fhe_long() {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [false; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [false; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let output = "D1F0303482061111";
+
+    let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+
+    let mut kreyvium = KreyviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+
+    let mut vec = Vec::<bool>::with_capacity(64);
+    while vec.len() < 64 {
+        let cipher_outputs = kreyvium.next_64();
+        for c in cipher_outputs {
+            vec.push(c.decrypt(&client_key))
+        }
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+use tfhe::shortint::prelude::*;
+
+#[test]
+fn kreyvium_test_shortint_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0; 128];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0; 128];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+    let output = "D1F0303482061111".to_string();
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &hl_client_key).unwrap();
+
+    let mut kreyvium = KreyviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    let trans_ciphered_message = kreyvium.trans_encrypt_64(ciphered_message);
+    let ciphered_message = trans_ciphered_message.decrypt(&hl_client_key);
+
+    let hexadecimal = get_hexagonal_string_from_u64(vec![ciphered_message]);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_clear_byte() {
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key_bytes = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key_bytes[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv_bytes = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv_bytes[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let output = "D1F0303482061111".to_string();
+
+    let mut kreyvium = KreyviumStreamByte::<u8>::new(key_bytes, iv_bytes);
+
+    let mut vec = Vec::<u8>::with_capacity(8);
+    while vec.len() < 8 {
+        let outputs = kreyvium.next_64();
+        for c in outputs {
+            vec.push(c)
+        }
+    }
+
+    let hexadecimal = get_hexagonal_string_from_bytes(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_byte_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .enable_function_evaluation_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key_bytes = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key_bytes[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv_bytes = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv_bytes[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let cipher_key = key_bytes.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let output = "D1F0303482061111".to_string();
+
+    let mut kreyvium = KreyviumStreamByte::<FheUint8>::new(cipher_key, iv_bytes, &server_key);
+
+    let mut vec = Vec::<u8>::with_capacity(8);
+    while vec.len() < 8 {
+        let cipher_outputs = kreyvium.next_64();
+        for c in cipher_outputs {
+            vec.push(c.decrypt(&client_key))
+        }
+    }
+
+    let hexadecimal = get_hexagonal_string_from_bytes(vec);
+    assert_eq!(output, hexadecimal);
+}
+
+#[test]
+fn kreyvium_test_fhe_byte_transciphering_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .enable_function_evaluation_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB000000000000".to_string();
+    let mut key = [0u8; 16];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC000000000000".to_string();
+    let mut iv = [0u8; 16];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let output = "D1F0303482061111".to_string();
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let ciphered_message = FheUint64::try_encrypt(0u64, &client_key).unwrap();
+
+    let mut kreyvium = KreyviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    let trans_ciphered_message = kreyvium.trans_encrypt_64(ciphered_message);
+    let ciphered_message = trans_ciphered_message.decrypt(&client_key);
+
+    let hexadecimal = get_hexagonal_string_from_u64(vec![ciphered_message]);
+    assert_eq!(output, hexadecimal);
+}

apps/trivium/src/lib.rs

@@ -0,0 +1,10 @@
+mod static_deque;
+
+mod kreyvium;
+pub use kreyvium::{KreyviumStream, KreyviumStreamByte, KreyviumStreamShortint};
+
+mod trivium;
+pub use trivium::{TriviumStream, TriviumStreamByte, TriviumStreamShortint};
+
+mod trans_ciphering;
+pub use trans_ciphering::TransCiphering;

apps/trivium/src/static_deque/mod.rs

@@ -0,0 +1,4 @@
+mod static_deque;
+pub use static_deque::StaticDeque;
+mod static_byte_deque;
+pub use static_byte_deque::{StaticByteDeque, StaticByteDequeInput};

apps/trivium/src/static_deque/static_byte_deque.rs

@@ -0,0 +1,141 @@
+//! This module implements the StaticByteDeque struct: a deque of bytes. The idea
+//! is that this is a wrapper around StaticDeque, but StaticByteDeque has an additional
+//! functionnality: it can construct the "intermediate" bytes, made of parts of other bytes.
+//! This is pretending to store bits, and allows accessing bits in chunks of 8 consecutive.
+
+use crate::static_deque::StaticDeque;
+
+use tfhe::FheUint8;
+
+/// Internal trait specifying which operations are needed by StaticByteDeque
+pub trait StaticByteDequeInput<OpOutput>:
+    Clone
+    + std::ops::Shr<u8, Output = OpOutput>
+    + std::ops::Shl<u8, Output = OpOutput>
+    + std::ops::BitOr<Output = OpOutput>
+{
+}
+impl StaticByteDequeInput<u8> for u8 {}
+impl StaticByteDequeInput<u8> for &u8 {}
+impl StaticByteDequeInput<FheUint8> for FheUint8 {}
+impl StaticByteDequeInput<FheUint8> for &FheUint8 {}
+
+/// Here T must represent a type covering a byte, like u8 or FheUint8.
+#[derive(Clone)]
+pub struct StaticByteDeque<const N: usize, T> {
+    deque: StaticDeque<N, T>,
+}
+
+impl<const N: usize, T> StaticByteDeque<N, T>
+where
+    T: StaticByteDequeInput<T>,
+    for<'a> &'a T: StaticByteDequeInput<T>,
+{
+    /// Constructor always uses a fully initialized array, the first element of
+    /// which is oldest, the last is newest
+    pub fn new(_arr: [T; N]) -> Self {
+        Self {
+            deque: StaticDeque::<N, T>::new(_arr),
+        }
+    }
+
+    /// Elements are pushed via a byte element (covering 8 underlying bits)
+    pub fn push(&mut self, val: T) {
+        self.deque.push(val)
+    }
+
+    /// computes n shift in a row
+    pub fn n_shifts(&mut self, n: usize) {
+        self.deque.n_shifts(n);
+    }
+
+    /// Getter for the internal memory
+    #[allow(dead_code)]
+    fn get_arr(&self) -> &[T; N] {
+        self.deque.get_arr()
+    }
+
+    /// This returns a byte full of zeros, except maybe a one
+    /// at the specified location, if it is present in the deque
+    #[allow(dead_code)]
+    fn bit(&self, i: usize) -> T
+    where
+        for<'a> &'a T: std::ops::BitAnd<u8, Output = T>,
+    {
+        let byte: &T = &self.deque[i / 8];
+        let bit_selector: u8 = 1u8 << (i % 8);
+        byte & bit_selector
+    }
+
+    /// This function reconstructs an intermediate byte if necessary
+    pub fn byte(&self, i: usize) -> T {
+        let byte: &T = &self.deque[i / 8];
+        let bit_idx: u8 = (i % 8) as u8;
+
+        if bit_idx == 0 {
+            return byte.clone();
+        }
+
+        let byte_next: &T = &self.deque[i / 8 + 1];
+        return (byte << bit_idx) | (byte_next >> (8 - bit_idx as u8));
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::static_deque::StaticByteDeque;
+
+    #[test]
+    fn byte_deque_test() {
+        let mut deque = StaticByteDeque::<3, u8>::new([2, 64, 128]);
+        deque.push(4);
+
+        // Youngest: 4
+        assert!(deque.bit(0) == 0);
+        assert!(deque.bit(1) == 0);
+        assert!(deque.bit(2) > 0);
+        assert!(deque.bit(3) == 0);
+        assert!(deque.bit(4) == 0);
+        assert!(deque.bit(5) == 0);
+        assert!(deque.bit(6) == 0);
+        assert!(deque.bit(7) == 0);
+
+        // second youngest: 128
+        assert!(deque.bit(8 + 0) == 0);
+        assert!(deque.bit(8 + 1) == 0);
+        assert!(deque.bit(8 + 2) == 0);
+        assert!(deque.bit(8 + 3) == 0);
+        assert!(deque.bit(8 + 4) == 0);
+        assert!(deque.bit(8 + 5) == 0);
+        assert!(deque.bit(8 + 6) == 0);
+        assert!(deque.bit(8 + 7) > 0);
+
+        // oldest: 64
+        assert!(deque.bit(16 + 0) == 0);
+        assert!(deque.bit(16 + 1) == 0);
+        assert!(deque.bit(16 + 2) == 0);
+        assert!(deque.bit(16 + 3) == 0);
+        assert!(deque.bit(16 + 4) == 0);
+        assert!(deque.bit(16 + 5) == 0);
+        assert!(deque.bit(16 + 6) > 0);
+        assert!(deque.bit(16 + 7) == 0);
+
+        assert_eq!(deque.byte(0), 4u8);
+        assert_eq!(deque.byte(1), 9u8);
+        assert_eq!(deque.byte(2), 18u8);
+        assert_eq!(deque.byte(3), 36u8);
+        assert_eq!(deque.byte(4), 72u8);
+        assert_eq!(deque.byte(5), 144u8);
+        assert_eq!(deque.byte(6), 32u8);
+        assert_eq!(deque.byte(7), 64u8);
+        assert_eq!(deque.byte(8), 128u8);
+        assert_eq!(deque.byte(9), 0u8);
+        assert_eq!(deque.byte(10), 1u8);
+        assert_eq!(deque.byte(11), 2u8);
+        assert_eq!(deque.byte(12), 4u8);
+        assert_eq!(deque.byte(13), 8u8);
+        assert_eq!(deque.byte(14), 16u8);
+        assert_eq!(deque.byte(15), 32u8);
+        assert_eq!(deque.byte(16), 64u8);
+    }
+}

apps/trivium/src/static_deque/static_deque.rs

@@ -0,0 +1,135 @@
+//! This module implements the StaticDeque struct: a deque utility whose size
+//! is known at compile time. Construction, push, and indexing are publicly
+//! available.
+
+use core::ops::{Index, IndexMut};
+
+/// StaticDeque: a struct implementing a deque whose size is known at compile time.
+/// It has 2 members: the static array conatining the data (never empty), and a cursor
+/// equal to the index of the oldest element (and the next one to be overwritten).
+#[derive(Clone)]
+pub struct StaticDeque<const N: usize, T> {
+    arr: [T; N],
+    cursor: usize,
+}
+
+impl<const N: usize, T> StaticDeque<N, T> {
+    /// Constructor always uses a fully initialized array, the first element of
+    /// which is oldest, the last is newest
+    pub fn new(_arr: [T; N]) -> Self {
+        Self {
+            arr: _arr,
+            cursor: 0,
+        }
+    }
+
+    /// Push a new element to the deque, overwriting the oldest at the same time.
+    pub fn push(&mut self, val: T) {
+        self.arr[self.cursor] = val;
+        self.shift();
+    }
+
+    /// Shift: equivalent to pushing the oldest element
+    pub fn shift(&mut self) {
+        self.n_shifts(1);
+    }
+
+    /// computes n shift in a row
+    pub fn n_shifts(&mut self, n: usize) {
+        self.cursor += n;
+        self.cursor %= N;
+    }
+
+    /// Getter for the internal memory
+    #[allow(dead_code)]
+    pub fn get_arr(&self) -> &[T; N] {
+        &self.arr
+    }
+}
+
+/// Index trait for the StaticDeque: 0 is the youngest element, N-1 is the oldest,
+/// and above N will panic.
+impl<const N: usize, T> Index<usize> for StaticDeque<N, T> {
+    type Output = T;
+
+    /// 0 is youngest
+    fn index(&self, i: usize) -> &T {
+        if i >= N {
+            panic!("Index {:?} too high for size {:?}", i, N);
+        }
+        &self.arr[(N + self.cursor - i - 1) % N]
+    }
+}
+/// IndexMut trait for the StaticDeque: 0 is the youngest element, N-1 is the oldest,
+/// and above N will panic.
+impl<const N: usize, T> IndexMut<usize> for StaticDeque<N, T> {
+    /// 0 is youngest
+    fn index_mut(&mut self, i: usize) -> &mut T {
+        if i >= N {
+            panic!("Index {:?} too high for size {:?}", i, N);
+        }
+        &mut self.arr[(N + self.cursor - i - 1) % N]
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::static_deque::StaticDeque;
+
+    #[test]
+    fn test_static_deque() {
+        let a = [1, 2, 3, 4, 5, 6];
+
+        let mut static_deque = StaticDeque::new(a);
+        for i in 7..11 {
+            static_deque.push(i);
+        }
+        assert_eq!(*static_deque.get_arr(), [7, 8, 9, 10, 5, 6]);
+
+        for i in 11..15 {
+            static_deque.push(i);
+        }
+        assert_eq!(*static_deque.get_arr(), [13, 14, 9, 10, 11, 12]);
+
+        assert_eq!(static_deque[0], 14);
+        assert_eq!(static_deque[1], 13);
+        assert_eq!(static_deque[2], 12);
+        assert_eq!(static_deque[3], 11);
+        assert_eq!(static_deque[4], 10);
+        assert_eq!(static_deque[5], 9);
+    }
+
+    #[test]
+    fn test_static_deque_indexmut() {
+        let a = [1, 2, 3, 4, 5, 6];
+
+        let mut static_deque = StaticDeque::new(a);
+        for i in 7..11 {
+            static_deque.push(i);
+        }
+        assert_eq!(*static_deque.get_arr(), [7, 8, 9, 10, 5, 6]);
+
+        for i in 11..15 {
+            static_deque.push(i);
+        }
+        assert_eq!(*static_deque.get_arr(), [13, 14, 9, 10, 11, 12]);
+
+        static_deque[1] = 100;
+
+        assert_eq!(static_deque[0], 14);
+        assert_eq!(static_deque[1], 100);
+        assert_eq!(static_deque[2], 12);
+        assert_eq!(static_deque[3], 11);
+        assert_eq!(static_deque[4], 10);
+        assert_eq!(static_deque[5], 9);
+    }
+
+    #[test]
+    #[should_panic]
+    fn test_static_deque_index_fail() {
+        let a = [1, 2, 3, 4, 5, 6];
+
+        let static_deque = StaticDeque::new(a);
+        let _ = static_deque[6];
+    }
+}

apps/trivium/src/trans_ciphering/mod.rs

@@ -0,0 +1,118 @@
+//! This module will contain extensions of some TriviumStream of KreyviumStream objects,
+//! when trans ciphering is available to them.
+
+use crate::{KreyviumStreamByte, KreyviumStreamShortint, TriviumStreamByte, TriviumStreamShortint};
+use tfhe::shortint::Ciphertext;
+
+use tfhe::{set_server_key, unset_server_key, FheUint64, FheUint8, ServerKey};
+
+use rayon::prelude::*;
+
+/// Triat specifying the interface for trans ciphering a FheUint64 object. Since it is meant
+/// to be used with stream ciphers, encryption and decryption are by default the same.
+pub trait TransCiphering {
+    fn trans_encrypt_64(&mut self, cipher: FheUint64) -> FheUint64;
+    fn trans_decrypt_64(&mut self, cipher: FheUint64) -> FheUint64 {
+        self.trans_encrypt_64(cipher)
+    }
+}
+
+fn transcipher_from_fheu8_stream(
+    stream: Vec<FheUint8>,
+    cipher: FheUint64,
+    fhe_server_key: &ServerKey,
+) -> FheUint64 {
+    assert_eq!(stream.len(), 8);
+
+    set_server_key(fhe_server_key.clone());
+    rayon::broadcast(|_| set_server_key(fhe_server_key.clone()));
+
+    let ret: FheUint64 = stream
+        .into_par_iter()
+        .enumerate()
+        .map(|(i, x)| &cipher ^ &(FheUint64::cast_from(x) << (8 * (7 - i) as u8)))
+        .reduce_with(|a, b| a | b)
+        .unwrap();
+
+    unset_server_key();
+    rayon::broadcast(|_| unset_server_key());
+
+    ret
+}
+
+fn transcipher_from_1_1_stream(
+    stream: Vec<Ciphertext>,
+    cipher: FheUint64,
+    hl_server_key: &ServerKey,
+    internal_server_key: &tfhe::shortint::ServerKey,
+    casting_key: &tfhe::shortint::KeySwitchingKey,
+) -> FheUint64 {
+    assert_eq!(stream.len(), 64);
+
+    let pairs = (0..32)
+        .into_par_iter()
+        .map(|i| {
+            let byte_idx = 7 - i / 4;
+            let pair_idx = i % 4;
+
+            let b0 = &stream[8 * byte_idx + 2 * pair_idx];
+            let b1 = &stream[8 * byte_idx + 2 * pair_idx + 1];
+
+            casting_key.cast(
+                &internal_server_key
+                    .unchecked_add(b0, &internal_server_key.unchecked_scalar_mul(b1, 2)),
+            )
+        })
+        .collect::<Vec<_>>();
+
+    set_server_key(hl_server_key.clone());
+    let ret = &cipher ^ &FheUint64::try_from(pairs).unwrap();
+    unset_server_key();
+    ret
+}
+
+impl TransCiphering for TriviumStreamByte<FheUint8> {
+    /// `TriviumStreamByte<FheUint8>`: since a full step outputs 8 bytes, these bytes
+    /// are each shifted by a number in [0, 8), and XORed with the input cipher
+    fn trans_encrypt_64(&mut self, cipher: FheUint64) -> FheUint64 {
+        transcipher_from_fheu8_stream(self.next_64(), cipher, self.get_server_key())
+    }
+}
+
+impl TransCiphering for KreyviumStreamByte<FheUint8> {
+    /// `KreyviumStreamByte<FheUint8>`: since a full step outputs 8 bytes, these bytes
+    /// are each shifted by a number in [0, 8), and XORed with the input cipher
+    fn trans_encrypt_64(&mut self, cipher: FheUint64) -> FheUint64 {
+        transcipher_from_fheu8_stream(self.next_64(), cipher, self.get_server_key())
+    }
+}
+
+impl TransCiphering for TriviumStreamShortint {
+    /// TriviumStreamShortint: since a full step outputs 64 shortints, these bits
+    /// are paired 2 by 2 in the HL parameter space and packed in a full word,
+    /// and XORed with the input cipher
+    fn trans_encrypt_64(&mut self, cipher: FheUint64) -> FheUint64 {
+        transcipher_from_1_1_stream(
+            self.next_64(),
+            cipher,
+            self.get_hl_server_key(),
+            self.get_internal_server_key(),
+            self.get_casting_key(),
+        )
+    }
+}
+
+impl TransCiphering for KreyviumStreamShortint {
+    /// KreyviumStreamShortint: since a full step outputs 64 shortints, these bits
+    /// are paired 2 by 2 in the HL parameter space and packed in a full word,
+    /// and XORed with the input cipher
+    fn trans_encrypt_64(&mut self, cipher: FheUint64) -> FheUint64 {
+        transcipher_from_1_1_stream(
+            self.next_64(),
+            cipher,
+            self.get_hl_server_key(),
+            self.get_internal_server_key(),
+            self.get_casting_key(),
+        )
+    }
+}

apps/trivium/src/trivium/mod.rs

@@ -0,0 +1,11 @@
+mod trivium;
+pub use trivium::TriviumStream;
+
+mod trivium_byte;
+pub use trivium_byte::TriviumStreamByte;
+
+mod trivium_shortint;
+pub use trivium_shortint::TriviumStreamShortint;
+
+#[cfg(test)]
+mod test;

apps/trivium/src/trivium/test.rs

@@ -0,0 +1,412 @@
+use tfhe::prelude::*;
+use tfhe::{generate_keys, ConfigBuilder, FheBool, FheUint64, FheUint8};
+
+use crate::{TransCiphering, TriviumStream, TriviumStreamByte, TriviumStreamShortint};
+
+// Values for these tests come from the github repo cantora/avr-crypto-lib, commit 2a5b018,
+// file testvectors/trivium-80.80.test-vectors
+
+fn get_hexadecimal_string_from_lsb_first_stream(a: Vec<bool>) -> String {
+    assert!(a.len() % 8 == 0);
+    let mut hexadecimal: String = "".to_string();
+    for test in a.chunks(8) {
+        // Encoding is bytes in LSB order
+        match test[4..8] {
+            [false, false, false, false] => hexadecimal.push('0'),
+            [true, false, false, false] => hexadecimal.push('1'),
+            [false, true, false, false] => hexadecimal.push('2'),
+            [true, true, false, false] => hexadecimal.push('3'),
+
+            [false, false, true, false] => hexadecimal.push('4'),
+            [true, false, true, false] => hexadecimal.push('5'),
+            [false, true, true, false] => hexadecimal.push('6'),
+            [true, true, true, false] => hexadecimal.push('7'),
+
+            [false, false, false, true] => hexadecimal.push('8'),
+            [true, false, false, true] => hexadecimal.push('9'),
+            [false, true, false, true] => hexadecimal.push('A'),
+            [true, true, false, true] => hexadecimal.push('B'),
+
+            [false, false, true, true] => hexadecimal.push('C'),
+            [true, false, true, true] => hexadecimal.push('D'),
+            [false, true, true, true] => hexadecimal.push('E'),
+            [true, true, true, true] => hexadecimal.push('F'),
+            _ => (),
+        };
+        match test[0..4] {
+            [false, false, false, false] => hexadecimal.push('0'),
+            [true, false, false, false] => hexadecimal.push('1'),
+            [false, true, false, false] => hexadecimal.push('2'),
+            [true, true, false, false] => hexadecimal.push('3'),
+
+            [false, false, true, false] => hexadecimal.push('4'),
+            [true, false, true, false] => hexadecimal.push('5'),
+            [false, true, true, false] => hexadecimal.push('6'),
+            [true, true, true, false] => hexadecimal.push('7'),
+
+            [false, false, false, true] => hexadecimal.push('8'),
+            [true, false, false, true] => hexadecimal.push('9'),
+            [false, true, false, true] => hexadecimal.push('A'),
+            [true, true, false, true] => hexadecimal.push('B'),
+
+            [false, false, true, true] => hexadecimal.push('C'),
+            [true, false, true, true] => hexadecimal.push('D'),
+            [false, true, true, true] => hexadecimal.push('E'),
+            [true, true, true, true] => hexadecimal.push('F'),
+            _ => (),
+        };
+    }
+    return hexadecimal;
+}
+
+fn get_hexagonal_string_from_bytes(a: Vec<u8>) -> String {
+    assert!(a.len() % 8 == 0);
+    let mut hexadecimal: String = "".to_string();
+    for test in a {
+        hexadecimal.push_str(&format!("{:02X?}", test));
+    }
+    return hexadecimal;
+}
+
+fn get_hexagonal_string_from_u64(a: Vec<u64>) -> String {
+    let mut hexadecimal: String = "".to_string();
+    for test in a {
+        hexadecimal.push_str(&format!("{:016X?}", test));
+    }
+    return hexadecimal;
+}
+
+#[test]
+fn trivium_test_1() {
+    let key = [false; 80];
+    let iv = [false; 80];
+    let output_0_63    = "FBE0BF265859051B517A2E4E239FC97F563203161907CF2DE7A8790FA1B2E9CDF75292030268B7382B4C1A759AA2599A285549986E74805903801A4CB5A5D4F2".to_string();
+    let output_192_255 = "0F1BE95091B8EA857B062AD52BADF47784AC6D9B2E3F85A9D79995043302F0FDF8B76E5BC8B7B4F0AA46CD20DDA04FDD197BC5E1635496828F2DBFB23F6BD5D0".to_string();
+    let output_256_319 = "80F9075437BAC73F696D0ABE3972F5FCE2192E5FCC13C0CB77D0ABA09126838D31A2D38A2087C46304C8A63B54109F679B0B1BC71E72A58D6DD3E0A3FF890D4A".to_string();
+    let output_448_511 = "68450EB0910A98EF1853E0FC1BED8AB6BB08DF5F167D34008C2A85284D4B886DD56883EE92BF18E69121670B4C81A5689C9B0538373D22EB923A28A2DB44C0EB".to_string();
+
+    let mut trivium = TriviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(512 * 8);
+    while vec.len() < 512 * 8 {
+        vec.push(trivium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+    assert_eq!(output_192_255, hexadecimal[192 * 2..256 * 2]);
+    assert_eq!(output_256_319, hexadecimal[256 * 2..320 * 2]);
+    assert_eq!(output_448_511, hexadecimal[448 * 2..512 * 2]);
+}
+
+#[test]
+fn trivium_test_2() {
+    let mut key = [false; 80];
+    let iv = [false; 80];
+    key[7] = true;
+
+    let output_0_63    = "38EB86FF730D7A9CAF8DF13A4420540DBB7B651464C87501552041C249F29A64D2FBF515610921EBE06C8F92CECF7F8098FF20CCCC6A62B97BE8EF7454FC80F9".to_string();
+    let output_192_255 = "EAF2625D411F61E41F6BAEEDDD5FE202600BD472F6C9CD1E9134A745D900EF6C023E4486538F09930CFD37157C0EB57C3EF6C954C42E707D52B743AD83CFF297".to_string();
+    let output_256_319 = "9A203CF7B2F3F09C43D188AA13A5A2021EE998C42F777E9B67C3FA221A0AA1B041AA9E86BC2F5C52AFF11F7D9EE480CB1187B20EB46D582743A52D7CD080A24A".to_string();
+    let output_448_511 = "EBF14772061C210843C18CEA2D2A275AE02FCB18E5D7942455FF77524E8A4CA51E369A847D1AEEFB9002FCD02342983CEAFA9D487CC2032B10192CD416310FA4".to_string();
+
+    let mut trivium = TriviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(512 * 8);
+    while vec.len() < 512 * 8 {
+        vec.push(trivium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+    assert_eq!(output_192_255, hexadecimal[192 * 2..256 * 2]);
+    assert_eq!(output_256_319, hexadecimal[256 * 2..320 * 2]);
+    assert_eq!(output_448_511, hexadecimal[448 * 2..512 * 2]);
+}
+
+#[test]
+fn trivium_test_3() {
+    let key = [false; 80];
+    let mut iv = [false; 80];
+    iv[7] = true;
+
+    let output_0_63    = "F8901736640549E3BA7D42EA2D07B9F49233C18D773008BD755585B1A8CBAB86C1E9A9B91F1AD33483FD6EE3696D659C9374260456A36AAE11F033A519CBD5D7".to_string();
+    let output_192_255 = "87423582AF64475C3A9C092E32A53C5FE07D35B4C9CA288A89A43DEF3913EA9237CA43342F3F8E83AD3A5C38D463516F94E3724455656A36279E3E924D442F06".to_string();
+    let output_256_319 = "D94389A90E6F3BF2BB4C8B057339AAD8AA2FEA238C29FCAC0D1FF1CB2535A07058BA995DD44CFC54CCEC54A5405B944C532D74E50EA370CDF1BA1CBAE93FC0B5".to_string();
+    let output_448_511 = "4844151714E56A3A2BBFBA426A1D60F9A4F265210A91EC29259AE2035234091C49FFB1893FA102D425C57C39EB4916F6D148DC83EBF7DE51EEB9ABFE045FB282".to_string();
+
+    let mut trivium = TriviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(512 * 8);
+    while vec.len() < 512 * 8 {
+        vec.push(trivium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+    assert_eq!(output_192_255, hexadecimal[192 * 2..256 * 2]);
+    assert_eq!(output_256_319, hexadecimal[256 * 2..320 * 2]);
+    assert_eq!(output_448_511, hexadecimal[448 * 2..512 * 2]);
+}
+
+#[test]
+fn trivium_test_4() {
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [false; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [false; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+    let output_65472_65535 = "C04C24A6938C8AF8A491D5E481271E0E601338F01067A86A795CA493AA4FF265619B8D448B706B7C88EE8395FC79E5B51AB40245BBF7773AE67DF86FCFB71F30".to_string();
+    let output_65536_65599 = "011A0D7EC32FA102C66C164CFCB189AED9F6982E8C7370A6A37414781192CEB155C534C1C8C9E53FDEADF2D3D0577DAD3A8EB2F6E5265F1E831C86844670BC69".to_string();
+    let output_131008_131071 = "48107374A9CE3AAF78221AE77789247CF6896A249ED75DCE0CF2D30EB9D889A0C61C9F480E5C07381DED9FAB2AD54333E82C89BA92E6E47FD828F1A66A8656E0".to_string();
+
+    let mut trivium = TriviumStream::<bool>::new(key, iv);
+
+    let mut vec = Vec::<bool>::with_capacity(131072 * 8);
+    while vec.len() < 131072 * 8 {
+        vec.push(trivium.next());
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+    assert_eq!(output_65472_65535, hexadecimal[65472 * 2..65536 * 2]);
+    assert_eq!(output_65536_65599, hexadecimal[65536 * 2..65600 * 2]);
+    assert_eq!(output_131008_131071, hexadecimal[131008 * 2..131072 * 2]);
+}
+
+#[test]
+fn trivium_test_clear_byte() {
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+    let output_65472_65535 = "C04C24A6938C8AF8A491D5E481271E0E601338F01067A86A795CA493AA4FF265619B8D448B706B7C88EE8395FC79E5B51AB40245BBF7773AE67DF86FCFB71F30".to_string();
+    let output_65536_65599 = "011A0D7EC32FA102C66C164CFCB189AED9F6982E8C7370A6A37414781192CEB155C534C1C8C9E53FDEADF2D3D0577DAD3A8EB2F6E5265F1E831C86844670BC69".to_string();
+    let output_131008_131071 = "48107374A9CE3AAF78221AE77789247CF6896A249ED75DCE0CF2D30EB9D889A0C61C9F480E5C07381DED9FAB2AD54333E82C89BA92E6E47FD828F1A66A8656E0".to_string();
+
+    let mut trivium = TriviumStreamByte::<u8>::new(key, iv);
+
+    let mut vec = Vec::<u8>::with_capacity(131072);
+    while vec.len() < 131072 {
+        let outputs = trivium.next_64();
+        for c in outputs {
+            vec.push(c)
+        }
+    }
+
+    let hexadecimal = get_hexagonal_string_from_bytes(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+    assert_eq!(output_65472_65535, hexadecimal[65472 * 2..65536 * 2]);
+    assert_eq!(output_65536_65599, hexadecimal[65536 * 2..65600 * 2]);
+    assert_eq!(output_131008_131071, hexadecimal[131008 * 2..131072 * 2]);
+}
+
+#[test]
+fn trivium_test_fhe_long() {
+    let config = ConfigBuilder::all_disabled().enable_default_bool().build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [false; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [false; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val: u8 = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2 == 1;
+            val >>= 1;
+        }
+    }
+
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+    let cipher_key = key.map(|x| FheBool::encrypt(x, &client_key));
+
+    let mut trivium = TriviumStream::<FheBool>::new(cipher_key, iv, &server_key);
+
+    let mut vec = Vec::<bool>::with_capacity(64 * 8);
+    while vec.len() < 64 * 8 {
+        let cipher_outputs = trivium.next_64();
+        for c in cipher_outputs {
+            vec.push(c.decrypt(&client_key))
+        }
+    }
+
+    let hexadecimal = get_hexadecimal_string_from_lsb_first_stream(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+}
+
+#[test]
+fn trivium_test_fhe_byte_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let mut trivium = TriviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    let mut vec = Vec::<u8>::with_capacity(64);
+    while vec.len() < 64 {
+        let cipher_outputs = trivium.next_64();
+        for c in cipher_outputs {
+            vec.push(c.decrypt(&client_key))
+        }
+    }
+
+    let hexadecimal = get_hexagonal_string_from_bytes(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+}
+
+#[test]
+fn trivium_test_fhe_byte_transciphering_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (client_key, server_key) = generate_keys(config);
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0u8; 10];
+
+    for i in (0..key_string.len()).step_by(2) {
+        key[i >> 1] = u8::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0u8; 10];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        iv[i >> 1] = u8::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+    }
+
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+    let cipher_key = key.map(|x| FheUint8::encrypt(x, &client_key));
+
+    let mut ciphered_message = vec![FheUint64::try_encrypt(0u64, &client_key).unwrap(); 9];
+
+    let mut trivium = TriviumStreamByte::<FheUint8>::new(cipher_key, iv, &server_key);
+
+    let mut vec = Vec::<u64>::with_capacity(8);
+    while vec.len() < 8 {
+        let trans_ciphered_message = trivium.trans_encrypt_64(ciphered_message.pop().unwrap());
+        vec.push(trans_ciphered_message.decrypt(&client_key));
+    }
+
+    let hexadecimal = get_hexagonal_string_from_u64(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+}
+
+use tfhe::shortint::prelude::*;
+
+#[test]
+fn trivium_test_shortint_long() {
+    let config = ConfigBuilder::all_disabled()
+        .enable_default_integers()
+        .build();
+    let (hl_client_key, hl_server_key) = generate_keys(config);
+    let underlying_ck: tfhe::shortint::ClientKey = (*hl_client_key.as_ref()).clone().into();
+    let underlying_sk: tfhe::shortint::ServerKey = (*hl_server_key.as_ref()).clone().into();
+
+    let (client_key, server_key): (ClientKey, ServerKey) = gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key, &server_key),
+        (&underlying_ck, &underlying_sk),
+        PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS,
+    );
+
+    let key_string = "0053A6F94C9FF24598EB".to_string();
+    let mut key = [0; 80];
+
+    for i in (0..key_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&key_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            key[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+
+    let iv_string = "0D74DB42A91077DE45AC".to_string();
+    let mut iv = [0; 80];
+
+    for i in (0..iv_string.len()).step_by(2) {
+        let mut val = u64::from_str_radix(&iv_string[i..i + 2], 16).unwrap();
+        for j in 0..8 {
+            iv[8 * (i >> 1) + j] = val % 2;
+            val >>= 1;
+        }
+    }
+    let output_0_63    = "F4CD954A717F26A7D6930830C4E7CF0819F80E03F25F342C64ADC66ABA7F8A8E6EAA49F23632AE3CD41A7BD290A0132F81C6D4043B6E397D7388F3A03B5FE358".to_string();
+
+    let cipher_key = key.map(|x| client_key.encrypt(x));
+
+    let mut ciphered_message = vec![FheUint64::try_encrypt(0u64, &hl_client_key).unwrap(); 9];
+
+    let mut trivium = TriviumStreamShortint::new(cipher_key, iv, server_key, ksk, hl_server_key);
+
+    let mut vec = Vec::<u64>::with_capacity(8);
+    while vec.len() < 8 {
+        let trans_ciphered_message = trivium.trans_encrypt_64(ciphered_message.pop().unwrap());
+        vec.push(trans_ciphered_message.decrypt(&hl_client_key));
+    }
+
+    let hexadecimal = get_hexagonal_string_from_u64(vec);
+    assert_eq!(output_0_63, hexadecimal[0..64 * 2]);
+}

apps/trivium/src/trivium/trivium.rs

@@ -0,0 +1,225 @@
+//! This module implements the Trivium stream cipher, using booleans or FheBool
+//! for the representaion of the inner bits.
+
+use crate::static_deque::StaticDeque;
+
+use tfhe::prelude::*;
+use tfhe::{set_server_key, unset_server_key, FheBool, ServerKey};
+
+use rayon::prelude::*;
+
+/// Internal trait specifying which operations are necessary for TriviumStream generic type
+pub trait TriviumBoolInput<OpOutput>:
+    Sized
+    + Clone
+    + std::ops::BitXor<Output = OpOutput>
+    + std::ops::BitAnd<Output = OpOutput>
+    + std::ops::Not<Output = OpOutput>
+{
+}
+impl TriviumBoolInput<bool> for bool {}
+impl TriviumBoolInput<bool> for &bool {}
+impl TriviumBoolInput<FheBool> for FheBool {}
+impl TriviumBoolInput<FheBool> for &FheBool {}
+
+/// TriviumStream: a struct implementing the Trivium stream cipher, using T for the internal
+/// representation of bits (bool or FheBool). To be able to compute FHE operations, it also owns
+/// an Option for a ServerKey.
+pub struct TriviumStream<T> {
+    a: StaticDeque<93, T>,
+    b: StaticDeque<84, T>,
+    c: StaticDeque<111, T>,
+    fhe_key: Option<ServerKey>,
+}
+
+impl TriviumStream<bool> {
+    /// Contructor for `TriviumStream<bool>`: arguments are the secret key and the input vector.
+    /// Outputs a TriviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(key: [bool; 80], iv: [bool; 80]) -> TriviumStream<bool> {
+        // Initialization of Trivium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register = [false; 93];
+        let mut b_register = [false; 84];
+        let mut c_register = [false; 111];
+
+        for i in 0..80 {
+            a_register[93 - 80 + i] = key[i];
+            b_register[84 - 80 + i] = iv[i];
+        }
+
+        c_register[0] = true;
+        c_register[1] = true;
+        c_register[2] = true;
+
+        TriviumStream::<bool>::new_from_registers(a_register, b_register, c_register, None)
+    }
+}
+
+impl TriviumStream<FheBool> {
+    /// Constructor for `TriviumStream<FheBool>`: arguments are the encrypted secret key and input
+    /// vector, and the FHE server key.
+    /// Outputs a TriviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(key: [FheBool; 80], iv: [bool; 80], sk: &ServerKey) -> TriviumStream<FheBool> {
+        set_server_key(sk.clone());
+
+        // Initialization of Trivium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register = [false; 93].map(|x| FheBool::encrypt_trivial(x));
+        let mut b_register = [false; 84].map(|x| FheBool::encrypt_trivial(x));
+        let mut c_register = [false; 111].map(|x| FheBool::encrypt_trivial(x));
+
+        for i in 0..80 {
+            a_register[93 - 80 + i] = key[i].clone();
+            b_register[84 - 80 + i] = FheBool::encrypt_trivial(iv[i]);
+        }
+
+        c_register[0] = FheBool::try_encrypt_trivial(true).unwrap();
+        c_register[1] = FheBool::try_encrypt_trivial(true).unwrap();
+        c_register[2] = FheBool::try_encrypt_trivial(true).unwrap();
+
+        unset_server_key();
+        TriviumStream::<FheBool>::new_from_registers(
+            a_register,
+            b_register,
+            c_register,
+            Some(sk.clone()),
+        )
+    }
+}
+
+impl<T> TriviumStream<T>
+where
+    T: TriviumBoolInput<T> + std::marker::Send + std::marker::Sync,
+    for<'a> &'a T: TriviumBoolInput<T>,
+{
+    /// Internal generic contructor: arguments are already prepared registers, and an optional FHE
+    /// server key
+    fn new_from_registers(
+        a_register: [T; 93],
+        b_register: [T; 84],
+        c_register: [T; 111],
+        key: Option<ServerKey>,
+    ) -> Self {
+        let mut ret = Self {
+            a: StaticDeque::<93, T>::new(a_register),
+            b: StaticDeque::<84, T>::new(b_register),
+            c: StaticDeque::<111, T>::new(c_register),
+            fhe_key: key,
+        };
+        ret.init();
+        ret
+    }
+
+    /// The specification of Trivium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes one turn of the stream, updating registers and outputting the new bit.
+    pub fn next(&mut self) -> T {
+        match &self.fhe_key {
+            Some(sk) => set_server_key(sk.clone()),
+            None => (),
+        };
+
+        let [o, a, b, c] = self.get_output_and_values(0);
+
+        self.a.push(a);
+        self.b.push(b);
+        self.c.push(c);
+
+        o
+    }
+
+    /// Computes a potential future step of Trivium, n terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, n: usize) -> [T; 4] {
+        assert!(n < 65);
+
+        let (((temp_a, temp_b), (temp_c, a_and)), (b_and, c_and)) = rayon::join(
+            || {
+                rayon::join(
+                    || {
+                        rayon::join(
+                            || &self.a[65 - n] ^ &self.a[92 - n],
+                            || &self.b[68 - n] ^ &self.b[83 - n],
+                        )
+                    },
+                    || {
+                        rayon::join(
+                            || &self.c[65 - n] ^ &self.c[110 - n],
+                            || &self.a[91 - n] & &self.a[90 - n],
+                        )
+                    },
+                )
+            },
+            || {
+                rayon::join(
+                    || &self.b[82 - n] & &self.b[81 - n],
+                    || &self.c[109 - n] & &self.c[108 - n],
+                )
+            },
+        );
+
+        let ((o, a), (b, c)) = rayon::join(
+            || {
+                rayon::join(
+                    || &(&temp_a ^ &temp_b) ^ &temp_c,
+                    || &temp_c ^ &(&c_and ^ &self.a[68 - n]),
+                )
+            },
+            || {
+                rayon::join(
+                    || &temp_a ^ &(&a_and ^ &self.b[77 - n]),
+                    || &temp_b ^ &(&b_and ^ &self.c[86 - n]),
+                )
+            },
+        );
+
+        [o, a, b, c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 64 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[T; 4]> {
+        (0..64)
+            .into_par_iter()
+            .map(|x| self.get_output_and_values(x))
+            .rev()
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<T> {
+        match &self.fhe_key {
+            Some(sk) => {
+                rayon::broadcast(|_| set_server_key(sk.clone()));
+            }
+            None => (),
+        }
+        let mut values = self.get_64_output_and_values();
+        match &self.fhe_key {
+            Some(_) => {
+                rayon::broadcast(|_| unset_server_key());
+            }
+            None => (),
+        }
+
+        let mut ret = Vec::<T>::with_capacity(64);
+
+        while let Some([o, a, b, c]) = values.pop() {
+            ret.push(o);
+            self.a.push(a);
+            self.b.push(b);
+            self.c.push(c);
+        }
+        ret
+    }
+}

apps/trivium/src/trivium/trivium_byte.rs

@@ -0,0 +1,241 @@
+//! This module implements the Trivium stream cipher, using u8 or FheUint8
+//! for the representaion of the inner bits.
+
+use crate::static_deque::{StaticByteDeque, StaticByteDequeInput};
+
+use tfhe::prelude::*;
+use tfhe::{set_server_key, unset_server_key, FheUint8, ServerKey};
+
+use rayon::prelude::*;
+
+/// Internal trait specifying which operations are necessary for TriviumStreamByte generic type
+pub trait TriviumByteInput<OpOutput>:
+    Sized
+    + Clone
+    + Send
+    + Sync
+    + StaticByteDequeInput<OpOutput>
+    + std::ops::BitXor<Output = OpOutput>
+    + std::ops::BitAnd<Output = OpOutput>
+    + std::ops::Shr<u8, Output = OpOutput>
+    + std::ops::Shl<u8, Output = OpOutput>
+    + std::ops::Add<Output = OpOutput>
+{
+}
+impl TriviumByteInput<u8> for u8 {}
+impl TriviumByteInput<u8> for &u8 {}
+impl TriviumByteInput<FheUint8> for FheUint8 {}
+impl TriviumByteInput<FheUint8> for &FheUint8 {}
+
+/// TriviumStreamByte: a struct implementing the Trivium stream cipher, using T for the internal
+/// representation of bits (u8 or FheUint8). To be able to compute FHE operations, it also owns
+/// an Option for a ServerKey.
+/// Since the original Trivium registers' sizes are not a multiple of 8, these registers (which
+/// store byte-like objects) have a size that is the eigth of the closest multiple of 8 above the
+/// originals' sizes.
+pub struct TriviumStreamByte<T> {
+    a_byte: StaticByteDeque<12, T>,
+    b_byte: StaticByteDeque<11, T>,
+    c_byte: StaticByteDeque<14, T>,
+    fhe_key: Option<ServerKey>,
+}
+
+impl TriviumStreamByte<u8> {
+    /// Contructor for `TriviumStreamByte<u8>`: arguments are the secret key and the input vector.
+    /// Outputs a TriviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(key: [u8; 10], iv: [u8; 10]) -> TriviumStreamByte<u8> {
+        // Initialization of Trivium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_byte_reg = [0u8; 12];
+        let mut b_byte_reg = [0u8; 11];
+        let mut c_byte_reg = [0u8; 14];
+
+        for i in 0..10 {
+            a_byte_reg[12 - 10 + i] = key[i];
+            b_byte_reg[11 - 10 + i] = iv[i];
+        }
+
+        // Magic number 14, aka 00001110: this represents the 3 ones at the beginning of the c
+        // registers, with additional zeros to make the register's size a multiple of 8.
+        c_byte_reg[0] = 14;
+
+        let mut ret =
+            TriviumStreamByte::<u8>::new_from_registers(a_byte_reg, b_byte_reg, c_byte_reg, None);
+        ret.init();
+        ret
+    }
+}
+
+impl TriviumStreamByte<FheUint8> {
+    /// Constructor for `TriviumStream<FheUint8>`: arguments are the encrypted secret key and input
+    /// vector, and the FHE server key.
+    /// Outputs a TriviumStream object already initialized (1152 steps have been run before
+    /// returning)
+    pub fn new(
+        key: [FheUint8; 10],
+        iv: [u8; 10],
+        server_key: &ServerKey,
+    ) -> TriviumStreamByte<FheUint8> {
+        set_server_key(server_key.clone());
+
+        // Initialization of Trivium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_byte_reg = [0u8; 12].map(|x| FheUint8::encrypt_trivial(x));
+        let mut b_byte_reg = [0u8; 11].map(|x| FheUint8::encrypt_trivial(x));
+        let mut c_byte_reg = [0u8; 14].map(|x| FheUint8::encrypt_trivial(x));
+
+        for i in 0..10 {
+            a_byte_reg[12 - 10 + i] = key[i].clone();
+            b_byte_reg[11 - 10 + i] = FheUint8::encrypt_trivial(iv[i]);
+        }
+
+        // Magic number 14, aka 00001110: this represents the 3 ones at the beginning of the c
+        // registers, with additional zeros to make the register's size a multiple of 8.
+        c_byte_reg[0] = FheUint8::encrypt_trivial(14u8);
+
+        unset_server_key();
+        let mut ret = TriviumStreamByte::<FheUint8>::new_from_registers(
+            a_byte_reg,
+            b_byte_reg,
+            c_byte_reg,
+            Some(server_key.clone()),
+        );
+        ret.init();
+        ret
+    }
+}
+
+impl<T> TriviumStreamByte<T>
+where
+    T: TriviumByteInput<T> + Send,
+    for<'a> &'a T: TriviumByteInput<T>,
+{
+    /// Internal generic contructor: arguments are already prepared registers, and an optional FHE
+    /// server key
+    fn new_from_registers(
+        a_register: [T; 12],
+        b_register: [T; 11],
+        c_register: [T; 14],
+        sk: Option<ServerKey>,
+    ) -> Self {
+        Self {
+            a_byte: StaticByteDeque::<12, T>::new(a_register),
+            b_byte: StaticByteDeque::<11, T>::new(b_register),
+            c_byte: StaticByteDeque::<14, T>::new(c_register),
+            fhe_key: sk,
+        }
+    }
+
+    /// The specification of Trivium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes 8 potential future step of Trivium, b*8 terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, b: usize) -> [T; 4] {
+        let n = b * 8 + 7;
+        assert!(n < 65);
+
+        let ((a1, a2, a3, a4, a5), ((b1, b2, b3, b4, b5), (c1, c2, c3, c4, c5))) = rayon::join(
+            || Self::get_bytes(&self.a_byte, [91 - n, 90 - n, 68 - n, 65 - n, 92 - n]),
+            || {
+                rayon::join(
+                    || Self::get_bytes(&self.b_byte, [82 - n, 81 - n, 77 - n, 68 - n, 83 - n]),
+                    || Self::get_bytes(&self.c_byte, [109 - n, 108 - n, 86 - n, 65 - n, 110 - n]),
+                )
+            },
+        );
+
+        let (((temp_a, temp_b), (temp_c, a_and)), (b_and, c_and)) = rayon::join(
+            || {
+                rayon::join(
+                    || rayon::join(|| a4 ^ a5, || b4 ^ b5),
+                    || rayon::join(|| c4 ^ c5, || a1 & a2),
+                )
+            },
+            || rayon::join(|| b1 & b2, || c1 & c2),
+        );
+
+        let (temp_a_2, temp_b_2, temp_c_2) = (temp_a.clone(), temp_b.clone(), temp_c.clone());
+
+        let ((o, a), (b, c)) = rayon::join(
+            || {
+                rayon::join(
+                    || (temp_a_2 ^ temp_b_2) ^ temp_c_2,
+                    || temp_c ^ ((c_and) ^ a3),
+                )
+            },
+            || rayon::join(|| temp_a ^ (a_and ^ b3), || temp_b ^ (b_and ^ c3)),
+        );
+
+        [o, a, b, c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 8 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[T; 4]> {
+        (0..8)
+            .into_par_iter()
+            .map(|i| self.get_output_and_values(i))
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits (in 8 bytes) all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<T> {
+        match &self.fhe_key {
+            Some(sk) => {
+                rayon::broadcast(|_| set_server_key(sk.clone()));
+            }
+            None => (),
+        }
+        let values = self.get_64_output_and_values();
+        match &self.fhe_key {
+            Some(_) => {
+                rayon::broadcast(|_| unset_server_key());
+            }
+            None => (),
+        }
+
+        let mut bytes = Vec::<T>::with_capacity(8);
+        for [o, a, b, c] in values {
+            self.a_byte.push(a);
+            self.b_byte.push(b);
+            self.c_byte.push(c);
+            bytes.push(o);
+        }
+
+        bytes
+    }
+
+    /// Reconstructs a bunch of 5 bytes in a parallel fashion.
+    fn get_bytes<const N: usize>(
+        reg: &StaticByteDeque<N, T>,
+        offsets: [usize; 5],
+    ) -> (T, T, T, T, T) {
+        let mut ret = offsets
+            .par_iter()
+            .rev()
+            .map(|&i| reg.byte(i))
+            .collect::<Vec<_>>();
+        (
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+            ret.pop().unwrap(),
+        )
+    }
+}
+
+impl TriviumStreamByte<FheUint8> {
+    pub fn get_server_key(&self) -> &ServerKey {
+        &self.fhe_key.as_ref().unwrap()
+    }
+}

apps/trivium/src/trivium/trivium_shortint.rs

@@ -0,0 +1,189 @@
+use crate::static_deque::StaticDeque;
+
+use tfhe::shortint::prelude::*;
+
+use rayon::prelude::*;
+
+/// TriviumStreamShortint: a struct implementing the Trivium stream cipher, using a generic
+/// Ciphertext for the internal representation of bits (intended to represent a single bit). To be
+/// able to compute FHE operations, it also owns a ServerKey.
+pub struct TriviumStreamShortint {
+    a: StaticDeque<93, Ciphertext>,
+    b: StaticDeque<84, Ciphertext>,
+    c: StaticDeque<111, Ciphertext>,
+    internal_server_key: ServerKey,
+    transciphering_casting_key: KeySwitchingKey,
+    hl_server_key: tfhe::ServerKey,
+}
+
+impl TriviumStreamShortint {
+    /// Contructor for TriviumStreamShortint: arguments are the secret key and the input vector, and
+    /// a ServerKey reference. Outputs a TriviumStream object already initialized (1152 steps
+    /// have been run before returning)
+    pub fn new(
+        key: [Ciphertext; 80],
+        iv: [u64; 80],
+        sk: ServerKey,
+        ksk: KeySwitchingKey,
+        hl_sk: tfhe::ServerKey,
+    ) -> Self {
+        // Initialization of Trivium registers: a has the secret key, b the input vector,
+        // and c a few ones.
+        let mut a_register: [Ciphertext; 93] = [0; 93].map(|x| sk.create_trivial(x));
+        let mut b_register: [Ciphertext; 84] = [0; 84].map(|x| sk.create_trivial(x));
+        let mut c_register: [Ciphertext; 111] = [0; 111].map(|x| sk.create_trivial(x));
+
+        for i in 0..80 {
+            a_register[93 - 80 + i] = key[i].clone();
+            b_register[84 - 80 + i] = sk.create_trivial(iv[i]);
+        }
+
+        c_register[0] = sk.create_trivial(1);
+        c_register[1] = sk.create_trivial(1);
+        c_register[2] = sk.create_trivial(1);
+
+        let mut ret = Self {
+            a: StaticDeque::<93, Ciphertext>::new(a_register),
+            b: StaticDeque::<84, Ciphertext>::new(b_register),
+            c: StaticDeque::<111, Ciphertext>::new(c_register),
+            internal_server_key: sk,
+            transciphering_casting_key: ksk,
+            hl_server_key: hl_sk,
+        };
+        ret.init();
+        ret
+    }
+
+    /// The specification of Trivium includes running 1152 (= 18*64) unused steps to mix up the
+    /// registers, before starting the proper stream
+    fn init(&mut self) {
+        for _ in 0..18 {
+            self.next_64();
+        }
+    }
+
+    /// Computes one turn of the stream, updating registers and outputting the new bit.
+    pub fn next(&mut self) -> Ciphertext {
+        let [o, a, b, c] = self.get_output_and_values(0);
+
+        self.a.push(a);
+        self.b.push(b);
+        self.c.push(c);
+
+        o
+    }
+
+    /// Computes a potential future step of Trivium, n terms in the future. This does not update
+    /// registers, but rather returns with the output, the three values that will be used to
+    /// update the registers, when the time is right. This function is meant to be used in
+    /// parallel.
+    fn get_output_and_values(&self, n: usize) -> [Ciphertext; 4] {
+        let (a1, a2, a3, a4, a5) = (
+            &self.a[65 - n],
+            &self.a[92 - n],
+            &self.a[91 - n],
+            &self.a[90 - n],
+            &self.a[68 - n],
+        );
+        let (b1, b2, b3, b4, b5) = (
+            &self.b[68 - n],
+            &self.b[83 - n],
+            &self.b[82 - n],
+            &self.b[81 - n],
+            &self.b[77 - n],
+        );
+        let (c1, c2, c3, c4, c5) = (
+            &self.c[65 - n],
+            &self.c[110 - n],
+            &self.c[109 - n],
+            &self.c[108 - n],
+            &self.c[86 - n],
+        );
+
+        let temp_a = self.internal_server_key.unchecked_add(a1, a2);
+        let temp_b = self.internal_server_key.unchecked_add(b1, b2);
+        let temp_c = self.internal_server_key.unchecked_add(c1, c2);
+
+        let ((new_a, new_b), (new_c, o)) = rayon::join(
+            || {
+                rayon::join(
+                    || {
+                        let mut new_a = self.internal_server_key.unchecked_bitand(c3, c4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_a, a5);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_a, &temp_c);
+                        self.internal_server_key.clear_carry_assign(&mut new_a);
+                        new_a
+                    },
+                    || {
+                        let mut new_b = self.internal_server_key.unchecked_bitand(a3, a4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_b, b5);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_b, &temp_a);
+                        self.internal_server_key.clear_carry_assign(&mut new_b);
+                        new_b
+                    },
+                )
+            },
+            || {
+                rayon::join(
+                    || {
+                        let mut new_c = self.internal_server_key.unchecked_bitand(b3, b4);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_c, c5);
+                        self.internal_server_key
+                            .unchecked_add_assign(&mut new_c, &temp_b);
+                        self.internal_server_key.clear_carry_assign(&mut new_c);
+                        new_c
+                    },
+                    || {
+                        self.internal_server_key.bitxor(
+                            &self.internal_server_key.unchecked_add(&temp_a, &temp_b),
+                            &temp_c,
+                        )
+                    },
+                )
+            },
+        );
+
+        [o, new_a, new_b, new_c]
+    }
+
+    /// This calls `get_output_and_values` in parallel 64 times, and stores all results in a Vec.
+    fn get_64_output_and_values(&self) -> Vec<[Ciphertext; 4]> {
+        (0..64)
+            .into_par_iter()
+            .map(|x| self.get_output_and_values(x))
+            .rev()
+            .collect()
+    }
+
+    /// Computes 64 turns of the stream, outputting the 64 bits all at once in a
+    /// Vec (first value is oldest, last is newest)
+    pub fn next_64(&mut self) -> Vec<Ciphertext> {
+        let mut values = self.get_64_output_and_values();
+
+        let mut ret = Vec::<Ciphertext>::with_capacity(64);
+        while let Some([o, a, b, c]) = values.pop() {
+            ret.push(o);
+            self.a.push(a);
+            self.b.push(b);
+            self.c.push(c);
+        }
+        ret
+    }
+
+    pub fn get_internal_server_key(&self) -> &ServerKey {
+        &self.internal_server_key
+    }
+
+    pub fn get_casting_key(&self) -> &KeySwitchingKey {
+        &self.transciphering_casting_key
+    }
+
+    pub fn get_hl_server_key(&self) -> &tfhe::ServerKey {
+        &self.hl_server_key
+    }
+}

ci/benchmark_parser.py

@@ -11,6 +11,8 @@ import json
 import sys
 
 
+ONE_HOUR_IN_NANOSECONDS = 3600E9
+
 parser = argparse.ArgumentParser()
 parser.add_argument('results',
                     help='Location of criterion benchmark results directory.'
@@ -36,10 +38,18 @@ parser.add_argument('--append-results', dest='append_results', action='store_tru
 parser.add_argument('--walk-subdirs', dest='walk_subdirs', action='store_true',
                     help='Check for results in subdirectories')
 parser.add_argument('--key-sizes', dest='key_sizes', action='store_true',
-                    help='Parse only the results regarding keys size measurments')
+                    help='Parse only the results regarding keys size measurements')
+parser.add_argument('--key-gen', dest='key_gen', action='store_true',
+                    help='Parse only the results regarding keys generation time measurements')
+parser.add_argument('--throughput', dest='throughput', action='store_true',
+                    help='Compute and append number of operations per second and'
+                         'operations per dollar')
+parser.add_argument('--backend', dest='backend', default='cpu',
+                    help='Backend on which benchmarks have run')
 
 
-def recursive_parse(directory, walk_subdirs=False, name_suffix=""):
+def recursive_parse(directory, walk_subdirs=False, name_suffix="", compute_throughput=False,
+                    hardware_hourly_cost=None):
     """
     Parse all the benchmark results in a directory. It will attempt to parse all the files having a
     .json extension at the top-level of this directory.
@@ -47,28 +57,100 @@ def recursive_parse(directory, walk_subdirs=False, name_suffix=""):
     :param directory: path to directory that contains raw results as :class:`pathlib.Path`
     :param walk_subdirs: traverse results subdirectories if parameters changes for benchmark case.
     :param name_suffix: a :class:`str` suffix to apply to each test name found
+    :param compute_throughput: compute number of operations per second and operations per
+        dollar
+    :param hardware_hourly_cost: hourly cost of the hardware used in dollar
 
-    :return: :class:`list` of data points
+    :return: tuple of :class:`list` as (data points, parsing failures)
     """
     excluded_directories = ["child_generate", "fork", "parent_generate", "report"]
-    result_values = list()
+    result_values = []
+    parsing_failures = []
+    bench_class = "evaluate"
+
     for dire in directory.iterdir():
         if dire.name in excluded_directories or not dire.is_dir():
             continue
         for subdir in dire.iterdir():
             if walk_subdirs:
-                subdir = subdir.joinpath("new")
-                if not subdir.exists():
-                    continue
+                if subdir.name == "new":
+                    pass
+                else:
+                    subdir = subdir.joinpath("new")
+                    if not subdir.exists():
+                        continue
             elif subdir.name != "new":
                 continue
 
-            test_name = parse_benchmark_file(subdir)
+            full_name, test_name = parse_benchmark_file(subdir)
+            if test_name is None:
+                parsing_failures.append((full_name, "'function_id' field is null in report"))
+                continue
+
+            try:
+                params, display_name, operator = get_parameters(test_name)
+            except Exception as err:
+                parsing_failures.append((full_name, f"failed to get parameters: {err}"))
+                continue
+
             for stat_name, value in parse_estimate_file(subdir).items():
                 test_name_parts = list(filter(None, [test_name, stat_name, name_suffix]))
-                result_values.append({"value": value, "test": "_".join(test_name_parts)})
 
-    return result_values
+                result_values.append(
+                    _create_point(
+                        value,
+                        "_".join(test_name_parts),
+                        bench_class,
+                        "latency",
+                        operator,
+                        params,
+                        display_name=display_name
+                    )
+                )
+
+                if stat_name == "mean" and compute_throughput:
+                    test_suffix = "ops-per-sec"
+                    test_name_parts.append(test_suffix)
+                    result_values.append(
+                        _create_point(
+                            compute_ops_per_second(value),
+                            "_".join(test_name_parts),
+                            bench_class,
+                            "throughput",
+                            operator,
+                            params,
+                            display_name="_".join([display_name, test_suffix])
+                        )
+                    )
+                    test_name_parts.pop()
+
+                    if hardware_hourly_cost is not None:
+                        test_suffix = "ops-per-dollar"
+                        test_name_parts.append(test_suffix)
+                        result_values.append(
+                            _create_point(
+                                compute_ops_per_dollar(value, hardware_hourly_cost),
+                                "_".join(test_name_parts),
+                                bench_class,
+                                "throughput",
+                                operator,
+                                params,
+                                display_name="_".join([display_name, test_suffix])
+                            )
+                        )
+
+    return result_values, parsing_failures
+
+
+def _create_point(value, test_name, bench_class, bench_type, operator, params, display_name=None):
+    return {
+        "value": value,
+        "test": test_name,
+        "name": display_name,
+        "class": bench_class,
+        "type": bench_type,
+        "operator": operator,
+        "params": params}
 
 
 def parse_benchmark_file(directory):
@@ -80,7 +162,7 @@ def parse_benchmark_file(directory):
     :return: name of the test as :class:`str`
     """
     raw_res = _parse_file_to_json(directory, "benchmark.json")
-    return raw_res["full_id"].replace(" ", "_")
+    return raw_res["full_id"], raw_res["function_id"]
 
 
 def parse_estimate_file(directory):
@@ -98,21 +180,102 @@ def parse_estimate_file(directory):
     }
 
 
+def _parse_key_results(result_file, bench_type):
+    """
+    Parse file containing results about operation on keys. The file must be formatted as CSV.
+
+    :param result_file: results file as :class:`pathlib.Path`
+
+    :return: tuple of :class:`list` as (data points, parsing failures)
+    """
+    result_values = []
+    parsing_failures = []
+
+    with result_file.open() as csv_file:
+        reader = csv.reader(csv_file)
+        for (test_name, value) in reader:
+            try:
+                params, display_name, operator = get_parameters(test_name)
+            except Exception as err:
+                parsing_failures.append((test_name, f"failed to get parameters: {err}"))
+                continue
+
+            result_values.append({
+                "value": int(value),
+                "test": test_name,
+                "name": display_name,
+                "class": "keygen",
+                "type": bench_type,
+                "operator": operator,
+                "params": params})
+
+    return result_values, parsing_failures
+
+
 def parse_key_sizes(result_file):
     """
     Parse file containing key sizes results. The file must be formatted as CSV.
 
     :param result_file: results file as :class:`pathlib.Path`
 
-    :return: :class:`list` of data points
+    :return: tuple of :class:`list` as (data points, parsing failures)
     """
-    result_values = list()
-    with result_file.open() as csv_file:
-        reader = csv.reader(csv_file)
-        for (test_name, value) in reader:
-            result_values.append({"value": int(value), "test": test_name})
+    return _parse_key_results(result_file, "keysize")
 
-    return result_values
+
+def parse_key_gen_time(result_file):
+    """
+    Parse file containing key generation time results. The file must be formatted as CSV.
+
+    :param result_file: results file as :class:`pathlib.Path`
+
+    :return: tuple of :class:`list` as (data points, parsing failures)
+    """
+    return _parse_key_results(result_file, "latency")
+
+
+def get_parameters(bench_id):
+    """
+    Get benchmarks parameters recorded for a given benchmark case.
+
+    :param bench_id: function name used for the benchmark case
+
+    :return: :class:`tuple` as ``(benchmark parameters, display name, operator type)``
+    """
+    params_dir = pathlib.Path("tfhe", "benchmarks_parameters", bench_id)
+    params = _parse_file_to_json(params_dir, "parameters.json")
+
+    display_name = params.pop("display_name")
+    operator = params.pop("operator_type")
+
+    # Put cryptographic parameters at the same level as the others parameters
+    crypto_params = params.pop("crypto_parameters")
+    params.update(crypto_params)
+
+    return params, display_name, operator
+
+
+def compute_ops_per_dollar(data_point, product_hourly_cost):
+    """
+    Compute numbers of operations per dollar for a given ``data_point``.
+
+    :param data_point: timing value measured during benchmark in nanoseconds
+    :param product_hourly_cost: cost in dollar per hour of hardware used
+
+    :return: number of operations per dollar
+    """
+    return ONE_HOUR_IN_NANOSECONDS / (product_hourly_cost * data_point)
+
+
+def compute_ops_per_second(data_point):
+    """
+    Compute numbers of operations per second for a given ``data_point``.
+
+    :param data_point: timing value measured during benchmark in nanoseconds
+
+    :return: number of operations per second
+    """
+    return 1E9 / data_point
 
 
 def _parse_file_to_json(directory, filename):
@@ -128,6 +291,9 @@ def dump_results(parsed_results, filename, input_args):
     :param filename: filename for dump file as :class:`pathlib.Path`
     :param input_args: CLI input arguments
     """
+    for point in parsed_results:
+        point["backend"] = input_args.backend
+
     if input_args.append_results:
         parsed_content = json.loads(filename.read_text())
         parsed_content["points"].extend(parsed_results)
@@ -156,10 +322,11 @@ def check_mandatory_args(input_args):
     if input_args.append_results:
         return
 
-    missing_args = list()
+    missing_args = []
     for arg_name in vars(input_args):
         if arg_name in ["results_dir", "output_file", "name_suffix",
-                        "append_results", "walk_subdirs", "key_sizes"]:
+                        "append_results", "walk_subdirs", "key_sizes",
+                        "key_gen", "throughput"]:
             continue
         if not getattr(input_args, arg_name):
             missing_args.append(arg_name)
@@ -174,13 +341,33 @@ if __name__ == "__main__":
     args = parser.parse_args()
     check_mandatory_args(args)
 
+    #failures = []
     raw_results = pathlib.Path(args.results)
-    if not args.key_sizes:
-        print("Parsing benchmark results... ")
-        results = recursive_parse(raw_results, args.walk_subdirs, args.name_suffix)
+    if args.key_sizes or args.key_gen:
+        if args.key_sizes:
+            print("Parsing key sizes results... ")
+            results, failures = parse_key_sizes(raw_results)
+
+        if args.key_gen:
+            print("Parsing key generation time results... ")
+            results, failures = parse_key_gen_time(raw_results)
     else:
-        print("Parsing key sizes results... ")
-        results = parse_key_sizes(raw_results)
+        print("Parsing benchmark results... ")
+        hardware_cost = None
+        if args.throughput:
+            print("Throughput computation enabled")
+            ec2_costs = json.loads(
+                pathlib.Path("ci/ec2_products_cost.json").read_text(encoding="utf-8"))
+            try:
+                hardware_cost = abs(ec2_costs[args.hardware])
+                print(f"Hardware hourly cost: {hardware_cost} $/h")
+            except KeyError:
+                print(f"Cannot find hardware hourly cost for '{args.hardware}'")
+                sys.exit(1)
+
+        results, failures = recursive_parse(raw_results, args.walk_subdirs, args.name_suffix,
+                                            args.throughput, hardware_cost)
+
     print("Parsing results done")
 
     output_file = pathlib.Path(args.output_file)
@@ -188,3 +375,10 @@ if __name__ == "__main__":
     dump_results(results, output_file, args)
 
     print("Done")
+
+    if failures:
+        print("\nParsing failed for some results")
+        print("-------------------------------")
+        for name, error in failures:
+            print(f"[{name}] {error}")
+        sys.exit(1)

ci/ec2_products_cost.json

@@ -0,0 +1,3 @@
+{
+  "m6i.metal": 7.168
+}

ci/parse_integer_benches_to_csv.py

@@ -0,0 +1,90 @@
+import argparse
+from pathlib import Path
+import json
+
+
+def main(args):
+    criterion_dir = Path(args.criterion_dir)
+    output_file = Path(args.output_file)
+
+    data = []
+    for json_file in sorted(criterion_dir.glob("**/*.json")):
+        if json_file.parent.name == "base" or json_file.name != "benchmark.json":
+            continue
+
+        try:
+            bench_data = json.loads(json_file.read_text())
+            estimate_file = json_file.with_name("estimates.json")
+            estimate_data = json.loads(estimate_file.read_text())
+
+            bench_function_id = bench_data["function_id"]
+
+            split = bench_function_id.split("::")
+            (_, function_name, parameter_set, bits) = split
+            (bits, _) = bits.split("_")
+            bits = int(bits)
+
+            estimate_mean_ms = estimate_data["mean"]["point_estimate"] / 1000000
+            estimate_lower_bound_ms = (
+                estimate_data["mean"]["confidence_interval"]["lower_bound"] / 1000000
+            )
+            estimate_upper_bound_ms = (
+                estimate_data["mean"]["confidence_interval"]["upper_bound"] / 1000000
+            )
+
+            data.append(
+                (
+                    function_name,
+                    parameter_set,
+                    bits,
+                    estimate_mean_ms,
+                    estimate_lower_bound_ms,
+                    estimate_upper_bound_ms,
+                )
+            )
+        except:
+            pass
+
+    if len(data) == 0:
+        print("No integer bench found, skipping writing output file")
+        return
+
+    with open(output_file, "w", encoding="utf-8") as output:
+        output.write(
+            "function_name,parameter_set,bits,mean_ms,"
+            "confidence_interval_lower_bound_ms,confidence_interval_upper_bound_ms\n"
+        )
+        # Sort by func_name, bit width and then parameters
+        data.sort(key=lambda x: (x[0], x[2], x[1]))
+
+        for dat in data:
+            (
+                function_name,
+                parameter_set,
+                bits,
+                estimate_mean_ms,
+                estimate_lower_bound_ms,
+                estimate_upper_bound_ms,
+            ) = dat
+            output.write(
+                f"{function_name},{parameter_set},{bits},{estimate_mean_ms},"
+                f"{estimate_lower_bound_ms},{estimate_upper_bound_ms}\n"
+            )
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser("Parse criterion results to csv file")
+    parser.add_argument(
+        "--criterion-dir",
+        type=str,
+        default="target/criterion",
+        help="Where to look for criterion result json files",
+    )
+    parser.add_argument(
+        "--output-file",
+        type=str,
+        default="parsed_benches.csv",
+        help="Path of the output file, will be csv formatted",
+    )
+
+    main(parser.parse_args())

ci/slab.toml

@@ -1,11 +1,16 @@
 [profile.cpu-big]
 region = "eu-west-3"
-image_id = "ami-06f8ee10713a8f809"
-instance_type = "c6i.8xlarge"
+image_id = "ami-0ab73f5bd11708a85"
+instance_type = "m6i.32xlarge"
+
+[profile.cpu-small]
+region = "eu-west-3"
+image_id = "ami-0ab73f5bd11708a85"
+instance_type = "m6i.4xlarge"
 
 [profile.bench]
 region = "eu-west-3"
-image_id = "ami-06f8ee10713a8f809"
+image_id = "ami-0ab73f5bd11708a85"
 instance_type = "m6i.metal"
 
 [command.cpu_test]
@@ -13,6 +18,36 @@ workflow = "aws_tfhe_tests.yml"
 profile = "cpu-big"
 check_run_name = "CPU AWS Tests"
 
+[command.cpu_integer_test]
+workflow = "aws_tfhe_integer_tests.yml"
+profile = "cpu-big"
+check_run_name = "CPU Integer AWS Tests"
+
+[command.cpu_multi_bit_test]
+workflow = "aws_tfhe_multi_bit_tests.yml"
+profile = "cpu-big"
+check_run_name = "CPU AWS Multi Bit Tests"
+
+[command.cpu_wasm_test]
+workflow = "aws_tfhe_wasm_tests.yml"
+profile = "cpu-small"
+check_run_name = "CPU AWS WASM Tests"
+
+[command.cpu_fast_test]
+workflow = "aws_tfhe_fast_tests.yml"
+profile = "cpu-big"
+check_run_name = "CPU AWS Fast Tests"
+
+[command.integer_bench]
+workflow = "integer_benchmark.yml"
+profile = "bench"
+check_run_name = "Integer CPU AWS Benchmarks"
+
+[command.integer_multi_bit_bench]
+workflow = "integer_multi_bit_benchmark.yml"
+profile = "bench"
+check_run_name = "Integer multi bit CPU AWS Benchmarks"
+
 [command.shortint_bench]
 workflow = "shortint_benchmark.yml"
 profile = "bench"
@@ -22,3 +57,13 @@ check_run_name = "Shortint CPU AWS Benchmarks"
 workflow = "boolean_benchmark.yml"
 profile = "bench"
 check_run_name = "Boolean CPU AWS Benchmarks"
+
+[command.pbs_bench]
+workflow = "pbs_benchmark.yml"
+profile = "bench"
+check_run_name = "PBS CPU AWS Benchmarks"
+
+[command.wasm_client_bench]
+workflow = "wasm_client_benchmark.yml"
+profile = "cpu-small"
+check_run_name = "WASM Client AWS Benchmarks"

docker/Dockerfile.wasm_tests

@@ -5,8 +5,8 @@ RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
 
 # Replace default archive.ubuntu.com with fr mirror
 # original archive showed performance issues and is farther away
-RUN sed -i 's|^deb http://archive|deb http://fr.archive|g' /etc/apt/sources.list && \
-    sed -i 's|^deb http://security|deb http://fr.archive|g' /etc/apt/sources.list
+RUN sed -i 's|^deb http://archive.ubuntu.com/ubuntu/|deb http://mirror.ubuntu.ikoula.com/|g' /etc/apt/sources.list && \
+    sed -i 's|^deb http://security.ubuntu.com/ubuntu/|deb http://mirror.ubuntu.ikoula.com/|g' /etc/apt/sources.list
 
 ENV CARGO_TARGET_DIR=/root/tfhe-rs-target
 
@@ -18,21 +18,22 @@ RUN apt-get update && \
     apt-get install -y \
     build-essential \
     curl \
-    git && \
+    git \
+    python3 \
+    python3-pip \
+    python3-venv && \
     rm -rf /var/lib/apt/lists/*
 
 RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs > install-rustup.sh && \
     chmod +x install-rustup.sh && \
     ./install-rustup.sh -y --default-toolchain "${RUST_TOOLCHAIN}" \
     -c rust-src -t wasm32-unknown-unknown && \
-    curl https://rustwasm.github.io/wasm-pack/installer/init.sh -sSf > install-wasm-pack.sh && \
-    chmod +x install-wasm-pack.sh && \
     . "$HOME/.cargo/env" && \
-    ./install-wasm-pack.sh -y && \
-    curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.2/install.sh > install-node.sh && \
+    cargo install wasm-pack && \
+    curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.3/install.sh > install-node.sh && \
     chmod +x install-node.sh && \
     ./install-node.sh && \
     . "$HOME/.nvm/nvm.sh" && \
-    bash -i -c 'nvm install node && nvm use node'
+    bash -i -c 'nvm install 20 && nvm use 20'
 
 WORKDIR /tfhe-wasm-tests/tfhe-rs/

scripts/c_api_tests.sh

@@ -40,7 +40,7 @@ mkdir -p "${TFHE_BUILD_DIR}"
 
 cd "${TFHE_BUILD_DIR}"
 
-cmake .. -DCMAKE_BUILD_TYPE=RELEASE
+cmake .. -DCMAKE_BUILD_TYPE=RELEASE -DCARGO_PROFILE="${CARGO_PROFILE}"
 
 make -j
 
@@ -48,4 +48,12 @@ if [[ "${BUILD_ONLY}" == "1" ]]; then
     exit 0
 fi
 
-make "test"
+nproc_bin=nproc
+
+# macOS detects CPUs differently
+if [[ $(uname) == "Darwin" ]]; then
+    nproc_bin="sysctl -n hw.logicalcpu"
+fi
+
+# Let's go parallel
+ARGS="-j$(${nproc_bin})" make test

scripts/check_cargo_min_ver.sh

@@ -2,18 +2,20 @@
 
 set -e
 
+CURR_DIR="$(dirname "$0")"
+REL_CARGO_TOML_PATH="${CURR_DIR}/../tfhe/Cargo.toml"
+MIN_RUST_VERSION="$(grep rust-version "${REL_CARGO_TOML_PATH}" | cut -d '=' -f 2 | xargs)"
+
 function usage() {
     echo "$0: check minimum cargo version"
     echo
     echo "--help                    Print this message"
     echo "--rust-toolchain          The toolchain to check the version for with leading"
-    echo "--min-rust-version        Check toolchain version is >= to this version, default is 1.65"
+    echo "--min-rust-version        Check toolchain version is >= to this version, default is ${MIN_RUST_VERSION}"
     echo
 }
 
 RUST_TOOLCHAIN=""
-# We set the default rust version 1.65 which is the minimum version required for stable GATs
-MIN_RUST_VERSION="1.65"
 
 while [ -n "$1" ]
 do

scripts/integer-tests.sh

@@ -0,0 +1,165 @@
+#!/bin/bash
+
+set -e
+
+function usage() {
+    echo "$0: shortint test runner"
+    echo
+    echo "--help                    Print this message"
+    echo "--rust-toolchain          The toolchain to run the tests with default: stable"
+    echo "--multi-bit               Run multi-bit tests only: default off"
+    echo "--cargo-profile           The cargo profile used to build tests"
+    echo
+}
+
+RUST_TOOLCHAIN="+stable"
+multi_bit=""
+not_multi_bit="_multi_bit"
+cargo_profile="release"
+
+while [ -n "$1" ]
+do
+   case "$1" in
+        "--help" | "-h" )
+            usage
+            exit 0
+            ;;
+
+        "--rust-toolchain" )
+            shift
+            RUST_TOOLCHAIN="$1"
+            ;;
+
+        "--multi-bit" )
+            multi_bit="_multi_bit"
+            not_multi_bit=""
+            ;;
+
+        "--cargo-profile" )
+            shift
+            cargo_profile="$1"
+            ;;
+
+        *)
+            echo "Unknown param : $1"
+            exit 1
+            ;;
+   esac
+   shift
+done
+
+if [[ "${RUST_TOOLCHAIN::1}" != "+" ]]; then
+    RUST_TOOLCHAIN="+${RUST_TOOLCHAIN}"
+fi
+
+CURR_DIR="$(dirname "$0")"
+ARCH_FEATURE="$("${CURR_DIR}/get_arch_feature.sh")"
+
+nproc_bin=nproc
+
+# macOS detects CPUs differently
+if [[ $(uname) == "Darwin" ]]; then
+    nproc_bin="sysctl -n hw.logicalcpu"
+fi
+
+n_threads="$(${nproc_bin})"
+
+if uname -a | grep "arm64"; then
+    if [[ $(uname) == "Darwin" ]]; then
+        # Keys are 4.7 gigs at max, CI M1 macs only has 8 gigs of RAM
+        n_threads=1
+    fi
+else
+    # Keys are 4.7 gigs at max, test machine has 32 gigs of RAM
+    n_threads=6
+fi
+
+if [[ "${BIG_TESTS_INSTANCE}" != TRUE ]]; then
+    if [[ "${FAST_TESTS}" != TRUE ]]; then
+        # block pbs are too slow for high params
+        # mul_crt_4_4 is extremely flaky (~80% failure)
+        # test_wopbs_bivariate_crt_wopbs_param_message generate tables that are too big at the moment
+        # test_integer_smart_mul_param_message_4_carry_4_ks_pbs is too slow
+        # so is test_integer_default_add_sequence_multi_thread_param_message_4_carry_4_ks_pbs
+        filter_expression="""\
+test(/^integer::.*${multi_bit}/) \
+${not_multi_bit:+"and not test(~${not_multi_bit})"} \
+and not test(/.*_block_pbs(_base)?_param_message_[34]_carry_[34]_ks_pbs$/) \
+and not test(~mul_crt_param_message_4_carry_4_ks_pbs) \
+and not test(/.*test_wopbs_bivariate_crt_wopbs_param_message_[34]_carry_[34]_ks_pbs$/) \
+and not test(/.*test_integer_smart_mul_param_message_4_carry_4_ks_pbs$/) \
+and not test(/.*test_integer_default_add_sequence_multi_thread_param_message_4_carry_4_ks_pbs$/)"""
+    else
+        # test only fast default operations with only two set of parameters
+        filter_expression="""\
+test(/^integer::.*${multi_bit}/) \
+${not_multi_bit:+"and not test(~${not_multi_bit})"} \
+and test(/.*_default_.*?_param${multi_bit}_message_[2-3]_carry_[2-3]${multi_bit:+"_group_2"}_ks_pbs/) \
+and not test(/.*_param_message_[14]_carry_[14]_ks_pbs$/) \
+and not test(/.*default_add_sequence_multi_thread_param_message_3_carry_3_ks_pbs$/)"""
+    fi
+
+    cargo "${RUST_TOOLCHAIN}" nextest run \
+        --tests \
+        --cargo-profile "${cargo_profile}" \
+        --package tfhe \
+        --profile ci \
+        --features="${ARCH_FEATURE}",integer,internal-keycache \
+        --test-threads "${n_threads}" \
+        -E "$filter_expression"
+
+    if [[ "${multi_bit}" == "" ]]; then
+        cargo "${RUST_TOOLCHAIN}" test \
+            --profile "${cargo_profile}" \
+            --package tfhe \
+            --features="${ARCH_FEATURE}",integer,internal-keycache \
+            --doc \
+            -- integer::
+    fi
+else
+    if [[ "${FAST_TESTS}" != TRUE ]]; then
+        # block pbs are too slow for high params
+        # mul_crt_4_4 is extremely flaky (~80% failure)
+        # test_wopbs_bivariate_crt_wopbs_param_message generate tables that are too big at the moment
+        # test_integer_smart_mul_param_message_4_carry_4_ks_pbs is too slow
+        # so is test_integer_default_add_sequence_multi_thread_param_message_4_carry_4_ks_pbs
+        filter_expression="""\
+test(/^integer::.*${multi_bit}/) \
+${not_multi_bit:+"and not test(~${not_multi_bit})"} \
+and not test(/.*_block_pbs(_base)?_param_message_[34]_carry_[34]_ks_pbs$/) \
+and not test(~mul_crt_param_message_4_carry_4_ks_pbs) \
+and not test(/.*test_wopbs_bivariate_crt_wopbs_param_message_[34]_carry_[34]_ks_pbs$/) \
+and not test(/.*test_integer_smart_mul_param_message_4_carry_4_ks_pbs$/) \
+and not test(/.*test_integer_default_add_sequence_multi_thread_param_message_4_carry_4_ks_pbs$/)"""
+    else
+        # test only fast default operations with only two set of parameters
+        filter_expression="""\
+test(/^integer::.*${multi_bit}/) \
+${not_multi_bit:+"and not test(~${not_multi_bit})"} \
+and test(/.*_default_.*?_param${multi_bit}_message_[2-3]_carry_[2-3]${multi_bit:+"_group_2"}_ks_pbs/) \
+and not test(/.*_param_message_[14]_carry_[14]_ks_pbs$/) \
+and not test(/.*default_add_sequence_multi_thread_param_message_3_carry_3_ks_pbs$/)"""
+    fi
+
+    num_cpu_threads="$(${nproc_bin})"
+    num_threads=$((num_cpu_threads * 2 / 3))
+    cargo "${RUST_TOOLCHAIN}" nextest run \
+        --tests \
+        --cargo-profile "${cargo_profile}" \
+        --package tfhe \
+        --profile ci \
+        --features="${ARCH_FEATURE}",integer,internal-keycache \
+        --test-threads $num_threads \
+        -E "$filter_expression"
+
+    if [[ "${multi_bit}" == "" ]]; then
+        cargo "${RUST_TOOLCHAIN}" test \
+            --profile "${cargo_profile}" \
+            --package tfhe \
+            --features="${ARCH_FEATURE}",integer,internal-keycache \
+            --doc \
+            -- --test-threads="$(${nproc_bin})" integer::
+    fi
+fi
+
+echo "Test ran in $SECONDS seconds"

scripts/no_dbg_calls.sh

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+set -e
+
+THIS_SCRIPT_NAME="$(basename "$0")"
+
+TMP_FILE="$(mktemp)"
+
+COUNT="$(git grep -rniI "dbg!" . | grep -v "${THIS_SCRIPT_NAME}" | \
+    tee "${TMP_FILE}" | wc -l | tr -d '[:space:]')"
+
+cat "${TMP_FILE}"
+rm -rf "${TMP_FILE}"
+
+if [[ "${COUNT}" == "0" ]]; then
+    exit 0
+else
+    echo "dbg macro calls detected, see output log above"
+    exit 1
+fi

scripts/no_tfhe_typo.sh

@@ -6,7 +6,7 @@ THIS_SCRIPT_NAME="$(basename "$0")"
 
 TMP_FILE="$(mktemp)"
 
-COUNT="$(git grep -rniI "thfe" . | grep -v "${THIS_SCRIPT_NAME}" | \
+COUNT="$(git grep -rniI "thfe\|tfhr\|thfr" . | grep -v "${THIS_SCRIPT_NAME}" | \
     tee "${TMP_FILE}" | wc -l | tr -d '[:space:]')"
 
 cat "${TMP_FILE}"

scripts/shortint-tests.sh

@@ -2,6 +2,54 @@
 
 set -e
 
+function usage() {
+    echo "$0: shortint test runner"
+    echo
+    echo "--help                    Print this message"
+    echo "--rust-toolchain          The toolchain to run the tests with default: stable"
+    echo "--multi-bit               Run multi-bit tests only: default off"
+    echo "--cargo-profile           The cargo profile used to build tests"
+    echo
+}
+
+RUST_TOOLCHAIN="+stable"
+multi_bit=""
+cargo_profile="release"
+
+while [ -n "$1" ]
+do
+   case "$1" in
+        "--help" | "-h" )
+            usage
+            exit 0
+            ;;
+
+        "--rust-toolchain" )
+            shift
+            RUST_TOOLCHAIN="$1"
+            ;;
+
+        "--multi-bit" )
+            multi_bit="_multi_bit"
+            ;;
+
+        "--cargo-profile" )
+            shift
+            cargo_profile="$1"
+            ;;
+
+        *)
+            echo "Unknown param : $1"
+            exit 1
+            ;;
+   esac
+   shift
+done
+
+if [[ "${RUST_TOOLCHAIN::1}" != "+" ]]; then
+    RUST_TOOLCHAIN="+${RUST_TOOLCHAIN}"
+fi
+
 CURR_DIR="$(dirname "$0")"
 ARCH_FEATURE="$("${CURR_DIR}/get_arch_feature.sh")"
 
@@ -30,54 +78,117 @@ else
     n_threads_big=13
 fi
 
-filter_expression_small_params=''\
-'('\
-'   test(/^shortint::.*_param_message_1_carry_1$/)'\
-'or test(/^shortint::.*_param_message_1_carry_2$/)'\
-'or test(/^shortint::.*_param_message_1_carry_3$/)'\
-'or test(/^shortint::.*_param_message_1_carry_4$/)'\
-'or test(/^shortint::.*_param_message_1_carry_5$/)'\
-'or test(/^shortint::.*_param_message_1_carry_6$/)'\
-'or test(/^shortint::.*_param_message_2_carry_1$/)'\
-'or test(/^shortint::.*_param_message_2_carry_2$/)'\
-'or test(/^shortint::.*_param_message_2_carry_3$/)'\
-'or test(/^shortint::.*_param_message_3_carry_1$/)'\
-'or test(/^shortint::.*_param_message_3_carry_2$/)'\
-'or test(/^shortint::.*_param_message_3_carry_3$/)'\
-')'\
-'and not test(~smart_add_and_mul)' # This test is too slow
+if [[ "${BIG_TESTS_INSTANCE}" != TRUE ]]; then
+    if [[ "${FAST_TESTS}" != TRUE ]]; then
+       filter_expression_small_params="""\
+(\
+   test(/^shortint::.*_param${multi_bit}_message_1_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_4${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_5${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_6${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+)\
+and not test(~smart_add_and_mul)""" # This test is too slow
+    else
+        filter_expression_small_params="""\
+(\
+   test(/^shortint::.*_param${multi_bit}_message_2_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+)\
+and not test(~smart_add_and_mul)""" # This test is too slow
+    fi
 
-# Run tests only no examples or benches with small params and more threads
-cargo ${1:+"${1}"} nextest run \
-    --tests \
-    --release \
-    --package tfhe \
-    --profile ci \
-    --features="${ARCH_FEATURE}",shortint,internal-keycache \
-    --test-threads "${n_threads_small}" \
-    -E "${filter_expression_small_params}"
+    # Run tests only no examples or benches with small params and more threads
+    cargo "${RUST_TOOLCHAIN}" nextest run \
+        --tests \
+        --cargo-profile "${cargo_profile}" \
+        --package tfhe \
+        --profile ci \
+        --features="${ARCH_FEATURE}",shortint,internal-keycache \
+        --test-threads "${n_threads_small}" \
+        -E "${filter_expression_small_params}"
 
-filter_expression_big_params=''\
-'('\
-'   test(/^shortint::.*_param_message_4_carry_4$/)'\
-')'\
-'and not test(~smart_add_and_mul)'
+    if [[ "${FAST_TESTS}" != TRUE ]]; then
+        filter_expression_big_params="""\
+(\
+   test(/^shortint::.*_param${multi_bit}_message_4_carry_4${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+) \
+and not test(~smart_add_and_mul)"""
 
-# Run tests only no examples or benches with big params and less threads
-cargo ${1:+"${1}"} nextest run \
-    --tests \
-    --release \
-    --package tfhe \
-    --profile ci \
-    --features="${ARCH_FEATURE}",shortint,internal-keycache \
-    --test-threads "${n_threads_big}" \
-    -E "${filter_expression_big_params}"
+    # Run tests only no examples or benches with big params and less threads
+    cargo "${RUST_TOOLCHAIN}" nextest run \
+        --tests \
+        --cargo-profile "${cargo_profile}" \
+        --package tfhe \
+        --profile ci \
+        --features="${ARCH_FEATURE}",shortint,internal-keycache \
+        --test-threads "${n_threads_big}" \
+        -E "${filter_expression_big_params}"
 
-cargo ${1:+"${1}"} test \
-    --release \
-    --package tfhe \
-    --features="${ARCH_FEATURE}",shortint,internal-keycache \
-    --doc \
-    shortint::
+        if [[ "${multi_bit}" == "" ]]; then
+            cargo "${RUST_TOOLCHAIN}" test \
+                --profile "${cargo_profile}" \
+                --package tfhe \
+                --features="${ARCH_FEATURE}",shortint,internal-keycache \
+                --doc \
+                -- shortint::
+        fi
+    fi
+else
+    if [[ "${FAST_TESTS}" != TRUE ]]; then
+        filter_expression="""\
+(\
+   test(/^shortint::.*_param${multi_bit}_message_1_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_4${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_5${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_1_carry_6${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_3_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_4_carry_4${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+)\
+and not test(~smart_add_and_mul)""" # This test is too slow
+    else
+        filter_expression="""\
+(\
+   test(/^shortint::.*_param${multi_bit}_message_2_carry_1${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_2${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+or test(/^shortint::.*_param${multi_bit}_message_2_carry_3${multi_bit:+"_group_[0-9]"}(_compact_pk)?_ks_pbs/) \
+)\
+and not test(~smart_add_and_mul)""" # This test is too slow
+    fi
+
+    # Run tests only no examples or benches with small params and more threads
+    cargo "${RUST_TOOLCHAIN}" nextest run \
+        --tests \
+        --cargo-profile "${cargo_profile}" \
+        --package tfhe \
+        --profile ci \
+        --features="${ARCH_FEATURE}",shortint,internal-keycache \
+        --test-threads "$(${nproc_bin})" \
+        -E "${filter_expression}"
+
+    if [[ "${multi_bit}" == "" ]]; then
+        cargo "${RUST_TOOLCHAIN}" test \
+            --profile "${cargo_profile}" \
+            --package tfhe \
+            --features="${ARCH_FEATURE}",shortint,internal-keycache \
+            --doc \
+            -- --test-threads="$(${nproc_bin})" shortint::
+    fi
+fi
 
 echo "Test ran in $SECONDS seconds"

tfhe/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tfhe"
-version = "0.1.8"
+version = "0.3.2"
 edition = "2021"
 readme = "../README.md"
 keywords = ["fully", "homomorphic", "encryption", "fhe", "cryptography"]
@@ -10,22 +10,37 @@ repository = "https://github.com/zama-ai/tfhe-rs"
 license = "BSD-3-Clause-Clear"
 description = "TFHE-rs is a fully homomorphic encryption (FHE) library that implements Zama's variant of TFHE."
 build = "build.rs"
-exclude = ["/docs/", "/c_api_tests/", "/CMakeLists.txt", "/js_on_wasm_tests/"]
-rust-version = "1.65"
+exclude = [
+    "/docs/",
+    "/c_api_tests/",
+    "/CMakeLists.txt",
+    "/js_on_wasm_tests/",
+    "/web_wasm_parallel_tests/",
+]
+rust-version = "1.67"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dev-dependencies]
 rand = "0.8.5"
 rand_distr = "0.4.3"
-kolmogorov_smirnov = "1.1.0"
 paste = "1.0.7"
 lazy_static = { version = "1.4.0" }
-criterion = "0.3.5"
+criterion = "0.4.0"
 doc-comment = "0.3.3"
+serde_json = "1.0.94"
+clap = { version = "=4.2.7", features = ["derive"] }
 # Used in user documentation
 bincode = "1.3.3"
 fs2 = { version = "0.4.3" }
+itertools = "0.10.5"
+num_cpus = "1.15"
+# For erf and normality test
+libm = "0.2.6"
+test-case = "3.1.0"
+combine = "4.6.6"
+env_logger = "0.10.0"
+log = "0.4.19"
 
 [build-dependencies]
 cbindgen = { version = "0.24.3", optional = true }
@@ -38,10 +53,11 @@ concrete-csprng = { version = "0.3.0", features = [
 lazy_static = { version = "1.4.0", optional = true }
 serde = { version = "1.0", features = ["derive"] }
 rayon = { version = "1.5.0" }
-bincode = { version = "1.3.3" }
-concrete-fft = { version = "0.1", features = ["serde"] }
+bincode = { version = "1.3.3", optional = true }
+concrete-fft = { version = "0.2.1", features = ["serde", "fft128"] }
+pulp = "0.11"
 aligned-vec = { version = "0.5", features = ["serde"] }
-dyn-stack = { version = "0.8" }
+dyn-stack = { version = "0.9" }
 once_cell = "1.13"
 paste = "1.0.7"
 fs2 = { version = "0.4.3", optional = true }
@@ -49,22 +65,33 @@ fs2 = { version = "0.4.3", optional = true }
 itertools = "0.10.5"
 
 # wasm deps
-wasm-bindgen = { version = "0.2.63", features = [
+wasm-bindgen = { version = "=0.2.86", features = [
     "serde-serialize",
 ], optional = true }
+# MSRV was bumped in a minor update, pin to still be able to build in CI
+bumpalo = { version = "=3.14" }
+wasm-bindgen-rayon = { version = "1.0", optional = true }
 js-sys = { version = "0.3", optional = true }
 console_error_panic_hook = { version = "0.1.7", optional = true }
 serde-wasm-bindgen = { version = "0.4", optional = true }
 getrandom = { version = "0.2.8", optional = true }
+bytemuck = "=1.14.1"
 
 [features]
 boolean = []
 shortint = []
-internal-keycache = ["lazy_static", "fs2"]
+integer = ["shortint"]
+internal-keycache = ["lazy_static", "fs2", "bincode"]
 
-__c_api = ["cbindgen"]
+# Experimental section
+experimental = []
+experimental-force_fft_algo_dif4 = []
+# End experimental section
+
+__c_api = ["cbindgen", "bincode"]
 boolean-c-api = ["boolean", "__c_api"]
 shortint-c-api = ["shortint", "__c_api"]
+high-level-c-api = ["boolean-c-api", "shortint-c-api", "integer", "__c_api"]
 
 __wasm_api = [
     "wasm-bindgen",
@@ -73,11 +100,15 @@ __wasm_api = [
     "serde-wasm-bindgen",
     "getrandom",
     "getrandom/js",
+    "bincode",
 ]
 boolean-client-js-wasm-api = ["boolean", "__wasm_api"]
 shortint-client-js-wasm-api = ["shortint", "__wasm_api"]
+integer-client-js-wasm-api = ["integer", "__wasm_api"]
+high-level-client-js-wasm-api = ["boolean", "shortint", "integer", "__wasm_api"]
+parallel-wasm-api = ["wasm-bindgen-rayon"]
 
-nightly-avx512 = ["concrete-fft/nightly"]
+nightly-avx512 = ["concrete-fft/nightly", "pulp/nightly"]
 
 # Enable the x86_64 specific accelerated implementation of the random generator for the default
 # backend
@@ -105,7 +136,7 @@ aarch64-unix = ["aarch64", "seeder_unix"]
 
 [package.metadata.docs.rs]
 # TODO: manage builds for docs.rs based on their documentation https://docs.rs/about
-features = ["x86_64-unix", "boolean", "shortint"]
+features = ["x86_64-unix", "boolean", "shortint", "integer"]
 rustdoc-args = ["--html-in-header", "katex-header.html"]
 
 ###########
@@ -114,6 +145,23 @@ rustdoc-args = ["--html-in-header", "katex-header.html"]
 #         #
 ###########
 
+[[bench]]
+name = "pbs-bench"
+path = "benches/core_crypto/pbs_bench.rs"
+harness = false
+required-features = ["boolean", "shortint", "internal-keycache"]
+
+[[bench]]
+name = "dev-bench"
+path = "benches/core_crypto/dev_bench.rs"
+harness = false
+required-features = ["experimental", "internal-keycache"]
+
+[[bench]]
+name = "pbs128-bench"
+path = "benches/core_crypto/pbs128_bench.rs"
+harness = false
+
 [[bench]]
 name = "boolean-bench"
 path = "benches/boolean/bench.rs"
@@ -126,20 +174,68 @@ path = "benches/shortint/bench.rs"
 harness = false
 required-features = ["shortint", "internal-keycache"]
 
+[[bench]]
+name = "integer-bench"
+path = "benches/integer/bench.rs"
+harness = false
+required-features = ["integer", "internal-keycache"]
+
+[[bench]]
+name = "keygen"
+path = "benches/keygen/bench.rs"
+harness = false
+required-features = ["shortint", "internal-keycache"]
+
+[[bench]]
+name = "utilities"
+path = "benches/utilities.rs"
+harness = false
+required-features = ["boolean", "shortint", "integer", "internal-keycache"]
+
+# Examples used as tools
+
+[[example]]
+name = "wasm_benchmarks_parser"
+path = "examples/utilities/wasm_benchmarks_parser.rs"
+required-features = ["shortint", "internal-keycache"]
+
 [[example]]
 name = "generates_test_keys"
+path = "examples/utilities/generates_test_keys.rs"
 required-features = ["shortint", "internal-keycache"]
 
 [[example]]
 name = "boolean_key_sizes"
+path = "examples/utilities/boolean_key_sizes.rs"
 required-features = ["boolean", "internal-keycache"]
 
 [[example]]
 name = "shortint_key_sizes"
+path = "examples/utilities/shortint_key_sizes.rs"
 required-features = ["shortint", "internal-keycache"]
 
+[[example]]
+name = "hlapi_compact_pk_ct_sizes"
+path = "examples/utilities/hlapi_compact_pk_ct_sizes.rs"
+required-features = ["integer", "internal-keycache"]
+
 [[example]]
 name = "micro_bench_and"
+path = "examples/utilities/micro_bench_and.rs"
+required-features = ["boolean"]
+
+# Real use-case examples
+
+[[example]]
+name = "dark_market"
+required-features = ["integer", "internal-keycache"]
+
+[[example]]
+name = "regex_engine"
+required-features = ["integer"]
+
+[[example]]
+name = "sha256_bool"
 required-features = ["boolean"]
 
 [lib]

tfhe/LICENSE

@@ -1,28 +1,28 @@
-BSD 3-Clause Clear License
-
-Copyright © 2022 ZAMA.
-All rights reserved.
-
-Redistribution and use in source and binary forms, with or without modification,
-are permitted provided that the following conditions are met:
-
-1. Redistributions of source code must retain the above copyright notice, this
-list of conditions and the following disclaimer.
-
-2. Redistributions in binary form must reproduce the above copyright notice, this
-list of conditions and the following disclaimer in the documentation and/or other
-materials provided with the distribution.
-
-3. Neither the name of ZAMA nor the names of its contributors may be used to endorse
-or promote products derived from this software without specific prior written permission.
-
-NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED BY THIS LICENSE.
-THIS SOFTWARE IS PROVIDED BY THE ZAMA AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
-IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
-MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
-ZAMA OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
-OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
-OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
-ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
-NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
-ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+BSD 3-Clause Clear License
+
+Copyright © 2023 ZAMA.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without modification,
+are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this
+list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this
+list of conditions and the following disclaimer in the documentation and/or other
+materials provided with the distribution.
+
+3. Neither the name of ZAMA nor the names of its contributors may be used to endorse
+or promote products derived from this software without specific prior written permission.
+
+NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED BY THIS LICENSE.
+THIS SOFTWARE IS PROVIDED BY THE ZAMA AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
+IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
+ZAMA OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
+OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

tfhe/benches/boolean/bench.rs

@@ -1,20 +1,50 @@
+#[path = "../utilities.rs"]
+mod utilities;
+use crate::utilities::{write_to_json, CryptoParametersRecord, OperatorType};
+
 use criterion::{black_box, criterion_group, criterion_main, Criterion};
 use tfhe::boolean::client_key::ClientKey;
-use tfhe::boolean::parameters::{BooleanParameters, DEFAULT_PARAMETERS, TFHE_LIB_PARAMETERS};
-use tfhe::boolean::prelude::BinaryBooleanGates;
+use tfhe::boolean::parameters::{
+    BooleanParameters, DEFAULT_PARAMETERS, PARAMETERS_ERROR_PROB_2_POW_MINUS_165,
+    PARAMETERS_ERROR_PROB_2_POW_MINUS_165_KS_PBS,
+};
+use tfhe::boolean::prelude::{BinaryBooleanGates, DEFAULT_PARAMETERS_KS_PBS, TFHE_LIB_PARAMETERS};
 use tfhe::boolean::server_key::ServerKey;
 
 criterion_group!(
     gates_benches,
     bench_default_parameters,
-    bench_tfhe_lib_parameters
+    bench_default_parameters_ks_pbs,
+    bench_low_prob_parameters,
+    bench_low_prob_parameters_ks_pbs,
+    bench_tfhe_lib_parameters,
 );
 
 criterion_main!(gates_benches);
 
+/// Helper function to write boolean benchmarks parameters to disk in JSON format.
+pub fn write_to_json_boolean<T: Into<CryptoParametersRecord<u32>>>(
+    bench_id: &str,
+    params: T,
+    params_alias: impl Into<String>,
+    display_name: impl Into<String>,
+) {
+    write_to_json(
+        bench_id,
+        params,
+        params_alias,
+        display_name,
+        &OperatorType::Atomic,
+        1,
+        vec![1],
+    );
+}
+
 // Put all `bench_function` in one place
 // so the keygen is only run once per parameters saving time.
-fn bench_gates(c: &mut Criterion, params: BooleanParameters, parameter_name: &str) {
+fn benchs(c: &mut Criterion, params: BooleanParameters, parameter_name: &str) {
+    let mut bench_group = c.benchmark_group("gates_benches");
+
     let cks = ClientKey::new(&params);
     let sks = ServerKey::new(&cks);
 
@@ -22,32 +52,59 @@ fn bench_gates(c: &mut Criterion, params: BooleanParameters, parameter_name: &st
     let ct2 = cks.encrypt(false);
     let ct3 = cks.encrypt(true);
 
-    let id = format!("AND gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.and(&ct1, &ct2))));
+    let id = format!("AND::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.and(&ct1, &ct2))));
+    write_to_json_boolean(&id, params, parameter_name, "and");
 
-    let id = format!("NAND gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.nand(&ct1, &ct2))));
+    let id = format!("NAND::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.nand(&ct1, &ct2))));
+    write_to_json_boolean(&id, params, parameter_name, "nand");
 
-    let id = format!("OR gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.or(&ct1, &ct2))));
+    let id = format!("OR::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.or(&ct1, &ct2))));
+    write_to_json_boolean(&id, params, parameter_name, "or");
 
-    let id = format!("XOR gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.xor(&ct1, &ct2))));
+    let id = format!("XOR::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.xor(&ct1, &ct2))));
+    write_to_json_boolean(&id, params, parameter_name, "xor");
 
-    let id = format!("XNOR gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.xnor(&ct1, &ct2))));
+    let id = format!("XNOR::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.xnor(&ct1, &ct2))));
+    write_to_json_boolean(&id, params, parameter_name, "xnor");
 
-    let id = format!("NOT gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.not(&ct1))));
+    let id = format!("NOT::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.not(&ct1))));
+    write_to_json_boolean(&id, params, parameter_name, "not");
 
-    let id = format!("MUX gate {parameter_name}");
-    c.bench_function(&id, |b| b.iter(|| black_box(sks.mux(&ct1, &ct2, &ct3))));
+    let id = format!("MUX::{parameter_name}");
+    bench_group.bench_function(&id, |b| b.iter(|| black_box(sks.mux(&ct1, &ct2, &ct3))));
+    write_to_json_boolean(&id, params, parameter_name, "mux");
 }
 
 fn bench_default_parameters(c: &mut Criterion) {
-    bench_gates(c, DEFAULT_PARAMETERS, "DEFAULT_PARAMETERS");
+    benchs(c, DEFAULT_PARAMETERS, "DEFAULT_PARAMETERS");
+}
+
+fn bench_default_parameters_ks_pbs(c: &mut Criterion) {
+    benchs(c, DEFAULT_PARAMETERS_KS_PBS, "DEFAULT_PARAMETERS_KS_PBS");
+}
+
+fn bench_low_prob_parameters(c: &mut Criterion) {
+    benchs(
+        c,
+        PARAMETERS_ERROR_PROB_2_POW_MINUS_165,
+        "PARAMETERS_ERROR_PROB_2_POW_MINUS_165_KS_PBS",
+    );
+}
+
+fn bench_low_prob_parameters_ks_pbs(c: &mut Criterion) {
+    benchs(
+        c,
+        PARAMETERS_ERROR_PROB_2_POW_MINUS_165_KS_PBS,
+        "PARAMETERS_ERROR_PROB_2_POW_MINUS_165_KS_PBS",
+    );
 }
 
 fn bench_tfhe_lib_parameters(c: &mut Criterion) {
-    bench_gates(c, TFHE_LIB_PARAMETERS, "TFHE_LIB_PARAMETERS");
+    benchs(c, TFHE_LIB_PARAMETERS, " TFHE_LIB_PARAMETERS");
 }

tfhe/benches/core_crypto/dev_bench.rs

@@ -0,0 +1,332 @@
+use criterion::{black_box, criterion_group, criterion_main, Criterion};
+use tfhe::core_crypto::prelude::*;
+
+criterion_group!(
+    boolean_like_pbs_group,
+    multi_bit_pbs::<u32>,
+    pbs::<u32>,
+    mem_optimized_pbs::<u32>
+);
+
+criterion_group!(
+    shortint_like_pbs_group,
+    multi_bit_pbs::<u64>,
+    pbs::<u64>,
+    mem_optimized_pbs::<u64>
+);
+
+criterion_main!(boolean_like_pbs_group, shortint_like_pbs_group);
+
+fn get_bench_params<Scalar: Numeric>() -> (
+    LweDimension,
+    StandardDev,
+    DecompositionBaseLog,
+    DecompositionLevelCount,
+    GlweDimension,
+    PolynomialSize,
+    LweBskGroupingFactor,
+    ThreadCount,
+) {
+    if Scalar::BITS == 64 {
+        (
+            LweDimension(742),
+            StandardDev(0.000007069849454709433),
+            DecompositionBaseLog(3),
+            DecompositionLevelCount(5),
+            GlweDimension(1),
+            PolynomialSize(1024),
+            LweBskGroupingFactor(2),
+            ThreadCount(5),
+        )
+    } else if Scalar::BITS == 32 {
+        (
+            LweDimension(778),
+            StandardDev(0.000003725679281679651),
+            DecompositionBaseLog(18),
+            DecompositionLevelCount(1),
+            GlweDimension(3),
+            PolynomialSize(512),
+            LweBskGroupingFactor(2),
+            ThreadCount(5),
+        )
+    } else {
+        unreachable!()
+    }
+}
+
+fn multi_bit_pbs<Scalar: UnsignedTorus + CastInto<usize> + CastFrom<usize> + Sync>(
+    c: &mut Criterion,
+) {
+    // DISCLAIMER: these toy example parameters are not guaranteed to be secure or yield correct
+    // computations
+    // Define parameters for LweBootstrapKey creation
+
+    let (
+        mut input_lwe_dimension,
+        lwe_modular_std_dev,
+        decomp_base_log,
+        decomp_level_count,
+        glwe_dimension,
+        polynomial_size,
+        grouping_factor,
+        thread_count,
+    ) = get_bench_params::<Scalar>();
+
+    let ciphertext_modulus = CiphertextModulus::new_native();
+
+    while input_lwe_dimension.0 % grouping_factor.0 != 0 {
+        input_lwe_dimension = LweDimension(input_lwe_dimension.0 + 1);
+    }
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    // Create the LweSecretKey
+    let input_lwe_secret_key =
+        allocate_and_generate_new_binary_lwe_secret_key(input_lwe_dimension, &mut secret_generator);
+    let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+        allocate_and_generate_new_binary_glwe_secret_key(
+            glwe_dimension,
+            polynomial_size,
+            &mut secret_generator,
+        );
+    let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+    let multi_bit_bsk = FourierLweMultiBitBootstrapKey::new(
+        input_lwe_dimension,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        decomp_base_log,
+        decomp_level_count,
+        grouping_factor,
+    );
+
+    // Allocate a new LweCiphertext and encrypt our plaintext
+    let lwe_ciphertext_in = allocate_and_encrypt_new_lwe_ciphertext(
+        &input_lwe_secret_key,
+        Plaintext(Scalar::ZERO),
+        lwe_modular_std_dev,
+        ciphertext_modulus,
+        &mut encryption_generator,
+    );
+
+    let accumulator = GlweCiphertext::new(
+        Scalar::ZERO,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        ciphertext_modulus,
+    );
+
+    // Allocate the LweCiphertext to store the result of the PBS
+    let mut out_pbs_ct = LweCiphertext::new(
+        Scalar::ZERO,
+        output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+        ciphertext_modulus,
+    );
+
+    let id = format!("Multi Bit PBS {}", Scalar::BITS);
+    #[allow(clippy::unit_arg)]
+    {
+        c.bench_function(&id, |b| {
+            b.iter(|| {
+                multi_bit_programmable_bootstrap_lwe_ciphertext(
+                    &lwe_ciphertext_in,
+                    &mut out_pbs_ct,
+                    &accumulator.as_view(),
+                    &multi_bit_bsk,
+                    thread_count,
+                );
+                black_box(&mut out_pbs_ct);
+            })
+        });
+    }
+}
+
+fn pbs<Scalar: UnsignedTorus + CastInto<usize>>(c: &mut Criterion) {
+    // DISCLAIMER: these toy example parameters are not guaranteed to be secure or yield correct
+    // computations
+    // Define parameters for LweBootstrapKey creation
+
+    let (
+        input_lwe_dimension,
+        lwe_modular_std_dev,
+        decomp_base_log,
+        decomp_level_count,
+        glwe_dimension,
+        polynomial_size,
+        _,
+        _,
+    ) = get_bench_params::<Scalar>();
+
+    let ciphertext_modulus = CiphertextModulus::new_native();
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    // Create the LweSecretKey
+    let input_lwe_secret_key =
+        allocate_and_generate_new_binary_lwe_secret_key(input_lwe_dimension, &mut secret_generator);
+    let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+        allocate_and_generate_new_binary_glwe_secret_key(
+            glwe_dimension,
+            polynomial_size,
+            &mut secret_generator,
+        );
+    let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+    // Create the empty bootstrapping key in the Fourier domain
+    let fourier_bsk = FourierLweBootstrapKey::new(
+        input_lwe_dimension,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        decomp_base_log,
+        decomp_level_count,
+    );
+
+    // Allocate a new LweCiphertext and encrypt our plaintext
+    let lwe_ciphertext_in = allocate_and_encrypt_new_lwe_ciphertext(
+        &input_lwe_secret_key,
+        Plaintext(Scalar::ZERO),
+        lwe_modular_std_dev,
+        ciphertext_modulus,
+        &mut encryption_generator,
+    );
+
+    let accumulator = GlweCiphertext::new(
+        Scalar::ZERO,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        ciphertext_modulus,
+    );
+
+    // Allocate the LweCiphertext to store the result of the PBS
+    let mut out_pbs_ct = LweCiphertext::new(
+        Scalar::ZERO,
+        output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+        ciphertext_modulus,
+    );
+
+    let id = format!("PBS {}", Scalar::BITS);
+    {
+        c.bench_function(&id, |b| {
+            b.iter(|| {
+                programmable_bootstrap_lwe_ciphertext(
+                    &lwe_ciphertext_in,
+                    &mut out_pbs_ct,
+                    &accumulator.as_view(),
+                    &fourier_bsk,
+                );
+                black_box(&mut out_pbs_ct);
+            })
+        });
+    }
+}
+
+fn mem_optimized_pbs<Scalar: UnsignedTorus + CastInto<usize>>(c: &mut Criterion) {
+    // DISCLAIMER: these toy example parameters are not guaranteed to be secure or yield correct
+    // computations
+    // Define parameters for LweBootstrapKey creation
+
+    let (
+        input_lwe_dimension,
+        lwe_modular_std_dev,
+        decomp_base_log,
+        decomp_level_count,
+        glwe_dimension,
+        polynomial_size,
+        _,
+        _,
+    ) = get_bench_params::<Scalar>();
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    // Create the LweSecretKey
+    let input_lwe_secret_key =
+        allocate_and_generate_new_binary_lwe_secret_key(input_lwe_dimension, &mut secret_generator);
+    let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+        allocate_and_generate_new_binary_glwe_secret_key(
+            glwe_dimension,
+            polynomial_size,
+            &mut secret_generator,
+        );
+    let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+    // Create the empty bootstrapping key in the Fourier domain
+    let fourier_bsk = FourierLweBootstrapKey::new(
+        input_lwe_dimension,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        decomp_base_log,
+        decomp_level_count,
+    );
+
+    // Allocate a new LweCiphertext and encrypt our plaintext
+    let lwe_ciphertext_in = allocate_and_encrypt_new_lwe_ciphertext(
+        &input_lwe_secret_key,
+        Plaintext(Scalar::ZERO),
+        lwe_modular_std_dev,
+        ciphertext_modulus,
+        &mut encryption_generator,
+    );
+
+    let accumulator = GlweCiphertext::new(
+        Scalar::ZERO,
+        glwe_dimension.to_glwe_size(),
+        polynomial_size,
+        ciphertext_modulus,
+    );
+    // Allocate the LweCiphertext to store the result of the PBS
+    let mut out_pbs_ct = LweCiphertext::new(
+        Scalar::ZERO,
+        output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+        ciphertext_modulus,
+    );
+
+    let mut buffers = ComputationBuffers::new();
+
+    let fft = Fft::new(fourier_bsk.polynomial_size());
+    let fft = fft.as_view();
+
+    buffers.resize(
+        programmable_bootstrap_lwe_ciphertext_mem_optimized_requirement::<Scalar>(
+            fourier_bsk.glwe_size(),
+            fourier_bsk.polynomial_size(),
+            fft,
+        )
+        .unwrap()
+        .unaligned_bytes_required(),
+    );
+
+    let id = format!("PBS mem-optimized {}", Scalar::BITS);
+    {
+        c.bench_function(&id, |b| {
+            b.iter(|| {
+                programmable_bootstrap_lwe_ciphertext_mem_optimized(
+                    &lwe_ciphertext_in,
+                    &mut out_pbs_ct,
+                    &accumulator.as_view(),
+                    &fourier_bsk,
+                    fft,
+                    buffers.stack(),
+                );
+                black_box(&mut out_pbs_ct);
+            })
+        });
+    }
+}

tfhe/benches/core_crypto/pbs128_bench.rs

@@ -0,0 +1,108 @@
+use criterion::{criterion_group, criterion_main, Criterion};
+use dyn_stack::PodStack;
+
+fn sqr(x: f64) -> f64 {
+    x * x
+}
+
+fn criterion_bench(c: &mut Criterion) {
+    {
+        use tfhe::core_crypto::fft_impl::fft128::crypto::bootstrap::bootstrap_scratch;
+        use tfhe::core_crypto::prelude::*;
+        type Scalar = u128;
+
+        let small_lwe_dimension = LweDimension(742);
+        let glwe_dimension = GlweDimension(1);
+        let polynomial_size = PolynomialSize(2048);
+        let lwe_modular_std_dev = StandardDev(sqr(0.000007069849454709433));
+        let pbs_base_log = DecompositionBaseLog(23);
+        let pbs_level = DecompositionLevelCount(1);
+        let ciphertext_modulus = CiphertextModulus::new_native();
+
+        let mut boxed_seeder = new_seeder();
+        let seeder = boxed_seeder.as_mut();
+
+        let mut secret_generator =
+            SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+        let mut encryption_generator =
+            EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+
+        let small_lwe_sk =
+            LweSecretKey::generate_new_binary(small_lwe_dimension, &mut secret_generator);
+
+        let glwe_sk = GlweSecretKey::<Vec<Scalar>>::generate_new_binary(
+            glwe_dimension,
+            polynomial_size,
+            &mut secret_generator,
+        );
+
+        let big_lwe_sk = glwe_sk.into_lwe_secret_key();
+
+        let fourier_bsk = Fourier128LweBootstrapKey::new(
+            small_lwe_dimension,
+            glwe_dimension.to_glwe_size(),
+            polynomial_size,
+            pbs_base_log,
+            pbs_level,
+        );
+
+        let fft = Fft128::new(polynomial_size);
+        let fft = fft.as_view();
+
+        let message_modulus: Scalar = 1 << 4;
+
+        let input_message: Scalar = 3;
+
+        let delta: Scalar = (1 << (Scalar::BITS - 1)) / message_modulus;
+
+        let plaintext = Plaintext(input_message * delta);
+
+        let lwe_ciphertext_in: LweCiphertextOwned<Scalar> = allocate_and_encrypt_new_lwe_ciphertext(
+            &small_lwe_sk,
+            plaintext,
+            lwe_modular_std_dev,
+            ciphertext_modulus,
+            &mut encryption_generator,
+        );
+
+        let accumulator: GlweCiphertextOwned<Scalar> = GlweCiphertextOwned::new(
+            Scalar::ONE,
+            glwe_dimension.to_glwe_size(),
+            polynomial_size,
+            ciphertext_modulus,
+        );
+
+        let mut pbs_out: LweCiphertext<Vec<Scalar>> = LweCiphertext::new(
+            0,
+            big_lwe_sk.lwe_dimension().to_lwe_size(),
+            ciphertext_modulus,
+        );
+
+        let mut buf = vec![
+            0u8;
+            bootstrap_scratch::<Scalar>(
+                fourier_bsk.glwe_size(),
+                fourier_bsk.polynomial_size(),
+                fft
+            )
+            .unwrap()
+            .unaligned_bytes_required()
+        ];
+
+        c.bench_function("pbs128", |b| {
+            b.iter(|| {
+                fourier_bsk.bootstrap(
+                    &mut pbs_out,
+                    &lwe_ciphertext_in,
+                    &accumulator,
+                    fft,
+                    PodStack::new(&mut buf),
+                )
+            });
+        });
+    }
+}
+
+criterion_group!(benches, criterion_bench);
+criterion_main!(benches);

tfhe/benches/core_crypto/pbs_bench.rs

@@ -0,0 +1,550 @@
+#[path = "../utilities.rs"]
+mod utilities;
+use crate::utilities::{write_to_json, CryptoParametersRecord, OperatorType};
+use rayon::prelude::*;
+
+use criterion::{black_box, criterion_group, criterion_main, Criterion};
+use serde::Serialize;
+use tfhe::boolean::parameters::{
+    BooleanParameters, DEFAULT_PARAMETERS, PARAMETERS_ERROR_PROB_2_POW_MINUS_165,
+};
+use tfhe::core_crypto::prelude::*;
+use tfhe::shortint::keycache::NamedParam;
+use tfhe::shortint::parameters::*;
+use tfhe::shortint::ClassicPBSParameters;
+
+const SHORTINT_BENCH_PARAMS: [ClassicPBSParameters; 15] = [
+    PARAM_MESSAGE_1_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_1_CARRY_1_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_1_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_2_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_2_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_4_KS_PBS,
+    PARAM_MESSAGE_5_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_6_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_7_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_8_CARRY_0_KS_PBS,
+];
+
+const BOOLEAN_BENCH_PARAMS: [(&str, BooleanParameters); 2] = [
+    ("BOOLEAN_DEFAULT_PARAMS", DEFAULT_PARAMETERS),
+    (
+        "BOOLEAN_TFHE_LIB_PARAMS",
+        PARAMETERS_ERROR_PROB_2_POW_MINUS_165,
+    ),
+];
+
+criterion_group!(
+    name = pbs_group;
+    config = Criterion::default().sample_size(2000);
+    targets = mem_optimized_pbs::<u64>, mem_optimized_pbs::<u32>
+);
+
+criterion_group!(
+    name = multi_bit_pbs_group;
+    config = Criterion::default().sample_size(2000);
+    targets =   multi_bit_pbs::<u64>,
+                multi_bit_pbs::<u32>,
+                multi_bit_deterministic_pbs::<u64>,
+                multi_bit_deterministic_pbs::<u32>,
+);
+
+criterion_group!(
+    name = pbs_throughput_group;
+    config = Criterion::default().sample_size(100);
+    targets = pbs_throughput::<u64>, pbs_throughput::<u32>
+);
+
+criterion_main!(pbs_group, multi_bit_pbs_group, pbs_throughput_group);
+
+fn benchmark_parameters<Scalar: UnsignedInteger>(
+) -> Vec<(&'static str, CryptoParametersRecord<Scalar>)> {
+    if Scalar::BITS == 64 {
+        SHORTINT_BENCH_PARAMS
+            .iter()
+            .map(|params| {
+                (
+                    params.name(),
+                    <ClassicPBSParameters as Into<PBSParameters>>::into(*params)
+                        .to_owned()
+                        .into(),
+                )
+            })
+            .collect()
+    } else if Scalar::BITS == 32 {
+        BOOLEAN_BENCH_PARAMS
+            .iter()
+            .map(|(name, params)| (*name, params.to_owned().into()))
+            .collect()
+    } else {
+        vec![]
+    }
+}
+
+fn throughput_benchmark_parameters<Scalar: UnsignedInteger>(
+) -> Vec<(&'static str, CryptoParametersRecord<Scalar>)> {
+    if Scalar::BITS == 64 {
+        vec![
+            PARAM_MESSAGE_1_CARRY_1_KS_PBS,
+            PARAM_MESSAGE_2_CARRY_2_KS_PBS,
+            PARAM_MESSAGE_3_CARRY_3_KS_PBS,
+        ]
+        .iter()
+        .map(|params| {
+            (
+                params.name(),
+                <ClassicPBSParameters as Into<PBSParameters>>::into(*params)
+                    .to_owned()
+                    .into(),
+            )
+        })
+        .collect()
+    } else if Scalar::BITS == 32 {
+        BOOLEAN_BENCH_PARAMS
+            .iter()
+            .map(|(name, params)| (*name, params.to_owned().into()))
+            .collect()
+    } else {
+        vec![]
+    }
+}
+
+fn multi_bit_benchmark_parameters<Scalar: UnsignedInteger + Default>() -> Vec<(
+    &'static str,
+    CryptoParametersRecord<Scalar>,
+    LweBskGroupingFactor,
+)> {
+    if Scalar::BITS == 64 {
+        vec![
+            PARAM_MULTI_BIT_MESSAGE_1_CARRY_1_GROUP_2_KS_PBS,
+            PARAM_MULTI_BIT_MESSAGE_2_CARRY_2_GROUP_2_KS_PBS,
+            PARAM_MULTI_BIT_MESSAGE_3_CARRY_3_GROUP_2_KS_PBS,
+            PARAM_MULTI_BIT_MESSAGE_1_CARRY_1_GROUP_3_KS_PBS,
+            PARAM_MULTI_BIT_MESSAGE_2_CARRY_2_GROUP_3_KS_PBS,
+            PARAM_MULTI_BIT_MESSAGE_3_CARRY_3_GROUP_3_KS_PBS,
+        ]
+        .iter()
+        .map(|params| {
+            (
+                params.name(),
+                <MultiBitPBSParameters as Into<PBSParameters>>::into(*params)
+                    .to_owned()
+                    .into(),
+                params.grouping_factor,
+            )
+        })
+        .collect()
+    } else {
+        // For now there are no parameters available to test multi bit PBS on 32 bits.
+        vec![]
+    }
+}
+
+fn mem_optimized_pbs<Scalar: UnsignedTorus + CastInto<usize> + Serialize>(c: &mut Criterion) {
+    let bench_name = "PBS_mem-optimized";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    for (name, params) in benchmark_parameters::<Scalar>().iter() {
+        // Create the LweSecretKey
+        let input_lwe_secret_key = allocate_and_generate_new_binary_lwe_secret_key(
+            params.lwe_dimension.unwrap(),
+            &mut secret_generator,
+        );
+        let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+            allocate_and_generate_new_binary_glwe_secret_key(
+                params.glwe_dimension.unwrap(),
+                params.polynomial_size.unwrap(),
+                &mut secret_generator,
+            );
+        let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+        // Create the empty bootstrapping key in the Fourier domain
+        let fourier_bsk = FourierLweBootstrapKey::new(
+            params.lwe_dimension.unwrap(),
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            params.pbs_base_log.unwrap(),
+            params.pbs_level.unwrap(),
+        );
+
+        // Allocate a new LweCiphertext and encrypt our plaintext
+        let lwe_ciphertext_in: LweCiphertextOwned<Scalar> = allocate_and_encrypt_new_lwe_ciphertext(
+            &input_lwe_secret_key,
+            Plaintext(Scalar::ZERO),
+            params.lwe_modular_std_dev.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+            &mut encryption_generator,
+        );
+
+        let accumulator = GlweCiphertext::new(
+            Scalar::ZERO,
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        // Allocate the LweCiphertext to store the result of the PBS
+        let mut out_pbs_ct = LweCiphertext::new(
+            Scalar::ZERO,
+            output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        let mut buffers = ComputationBuffers::new();
+
+        let fft = Fft::new(fourier_bsk.polynomial_size());
+        let fft = fft.as_view();
+
+        buffers.resize(
+            programmable_bootstrap_lwe_ciphertext_mem_optimized_requirement::<Scalar>(
+                fourier_bsk.glwe_size(),
+                fourier_bsk.polynomial_size(),
+                fft,
+            )
+            .unwrap()
+            .unaligned_bytes_required(),
+        );
+
+        let id = format!("{bench_name}_{name}");
+        {
+            bench_group.bench_function(&id, |b| {
+                b.iter(|| {
+                    programmable_bootstrap_lwe_ciphertext_mem_optimized(
+                        &lwe_ciphertext_in,
+                        &mut out_pbs_ct,
+                        &accumulator.as_view(),
+                        &fourier_bsk,
+                        fft,
+                        buffers.stack(),
+                    );
+                    black_box(&mut out_pbs_ct);
+                })
+            });
+        }
+
+        let bit_size = (params.message_modulus.unwrap_or(2) as u32).ilog2();
+        write_to_json(
+            &id,
+            *params,
+            *name,
+            "pbs",
+            &OperatorType::Atomic,
+            bit_size,
+            vec![bit_size],
+        );
+    }
+}
+
+fn multi_bit_pbs<
+    Scalar: UnsignedTorus + CastInto<usize> + CastFrom<usize> + Default + Sync + Serialize,
+>(
+    c: &mut Criterion,
+) {
+    let bench_name = "multi_bits_PBS";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    for (name, params, grouping_factor) in multi_bit_benchmark_parameters::<Scalar>().iter() {
+        // Create the LweSecretKey
+        let input_lwe_secret_key = allocate_and_generate_new_binary_lwe_secret_key(
+            params.lwe_dimension.unwrap(),
+            &mut secret_generator,
+        );
+        let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+            allocate_and_generate_new_binary_glwe_secret_key(
+                params.glwe_dimension.unwrap(),
+                params.polynomial_size.unwrap(),
+                &mut secret_generator,
+            );
+        let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+        let multi_bit_bsk = FourierLweMultiBitBootstrapKey::new(
+            params.lwe_dimension.unwrap(),
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            params.pbs_base_log.unwrap(),
+            params.pbs_level.unwrap(),
+            *grouping_factor,
+        );
+
+        // Allocate a new LweCiphertext and encrypt our plaintext
+        let lwe_ciphertext_in = allocate_and_encrypt_new_lwe_ciphertext(
+            &input_lwe_secret_key,
+            Plaintext(Scalar::ZERO),
+            params.lwe_modular_std_dev.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+            &mut encryption_generator,
+        );
+
+        let accumulator = GlweCiphertext::new(
+            Scalar::ZERO,
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        // Allocate the LweCiphertext to store the result of the PBS
+        let mut out_pbs_ct = LweCiphertext::new(
+            Scalar::ZERO,
+            output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        let id = format!("{bench_name}_{name}_parallelized");
+        bench_group.bench_function(&id, |b| {
+            b.iter(|| {
+                multi_bit_programmable_bootstrap_lwe_ciphertext(
+                    &lwe_ciphertext_in,
+                    &mut out_pbs_ct,
+                    &accumulator.as_view(),
+                    &multi_bit_bsk,
+                    ThreadCount(10),
+                );
+                black_box(&mut out_pbs_ct);
+            })
+        });
+
+        let bit_size = params.message_modulus.unwrap().ilog2();
+        write_to_json(
+            &id,
+            *params,
+            *name,
+            "pbs",
+            &OperatorType::Atomic,
+            bit_size,
+            vec![bit_size],
+        );
+    }
+}
+
+fn multi_bit_deterministic_pbs<
+    Scalar: UnsignedTorus + CastInto<usize> + CastFrom<usize> + Default + Serialize + Sync,
+>(
+    c: &mut Criterion,
+) {
+    let bench_name = "multi_bits_deterministic_PBS";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    for (name, params, grouping_factor) in multi_bit_benchmark_parameters::<Scalar>().iter() {
+        // Create the LweSecretKey
+        let input_lwe_secret_key = allocate_and_generate_new_binary_lwe_secret_key(
+            params.lwe_dimension.unwrap(),
+            &mut secret_generator,
+        );
+        let output_glwe_secret_key: GlweSecretKeyOwned<Scalar> =
+            allocate_and_generate_new_binary_glwe_secret_key(
+                params.glwe_dimension.unwrap(),
+                params.polynomial_size.unwrap(),
+                &mut secret_generator,
+            );
+        let output_lwe_secret_key = output_glwe_secret_key.into_lwe_secret_key();
+
+        let multi_bit_bsk = FourierLweMultiBitBootstrapKey::new(
+            params.lwe_dimension.unwrap(),
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            params.pbs_base_log.unwrap(),
+            params.pbs_level.unwrap(),
+            *grouping_factor,
+        );
+
+        // Allocate a new LweCiphertext and encrypt our plaintext
+        let lwe_ciphertext_in = allocate_and_encrypt_new_lwe_ciphertext(
+            &input_lwe_secret_key,
+            Plaintext(Scalar::ZERO),
+            params.lwe_modular_std_dev.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+            &mut encryption_generator,
+        );
+
+        let accumulator = GlweCiphertext::new(
+            Scalar::ZERO,
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        // Allocate the LweCiphertext to store the result of the PBS
+        let mut out_pbs_ct = LweCiphertext::new(
+            Scalar::ZERO,
+            output_lwe_secret_key.lwe_dimension().to_lwe_size(),
+            tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+        );
+
+        let id = format!("{bench_name}_{name}_parallelized");
+        bench_group.bench_function(&id, |b| {
+            b.iter(|| {
+                multi_bit_deterministic_programmable_bootstrap_lwe_ciphertext(
+                    &lwe_ciphertext_in,
+                    &mut out_pbs_ct,
+                    &accumulator.as_view(),
+                    &multi_bit_bsk,
+                    ThreadCount(10),
+                );
+                black_box(&mut out_pbs_ct);
+            })
+        });
+
+        let bit_size = params.message_modulus.unwrap().ilog2();
+        write_to_json(
+            &id,
+            *params,
+            *name,
+            "pbs",
+            &OperatorType::Atomic,
+            bit_size,
+            vec![bit_size],
+        );
+    }
+}
+
+fn pbs_throughput<Scalar: UnsignedTorus + CastInto<usize> + Sync + Send + Serialize>(
+    c: &mut Criterion,
+) {
+    let bench_name = "PBS_throughput";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    // Create the PRNG
+    let mut seeder = new_seeder();
+    let seeder = seeder.as_mut();
+    let mut encryption_generator =
+        EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed(), seeder);
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(seeder.seed());
+
+    for (name, params) in throughput_benchmark_parameters::<Scalar>().iter() {
+        let input_lwe_secret_key = allocate_and_generate_new_binary_lwe_secret_key(
+            params.lwe_dimension.unwrap(),
+            &mut secret_generator,
+        );
+
+        let glwe_secret_key = GlweSecretKey::new_empty_key(
+            Scalar::ZERO,
+            params.glwe_dimension.unwrap(),
+            params.polynomial_size.unwrap(),
+        );
+        let big_lwe_sk = glwe_secret_key.into_lwe_secret_key();
+        let big_lwe_dimension = big_lwe_sk.lwe_dimension();
+
+        const NUM_CTS: usize = 512;
+        let lwe_vec: Vec<_> = (0..NUM_CTS)
+            .map(|_| {
+                allocate_and_encrypt_new_lwe_ciphertext(
+                    &input_lwe_secret_key,
+                    Plaintext(Scalar::ZERO),
+                    params.lwe_modular_std_dev.unwrap(),
+                    tfhe::core_crypto::prelude::CiphertextModulus::new_native(),
+                    &mut encryption_generator,
+                )
+            })
+            .collect();
+
+        let mut output_lwe_list = LweCiphertextList::new(
+            Scalar::ZERO,
+            big_lwe_dimension.to_lwe_size(),
+            LweCiphertextCount(NUM_CTS),
+            params.ciphertext_modulus.unwrap(),
+        );
+
+        let lwe_vec = lwe_vec;
+
+        let fft = Fft::new(params.polynomial_size.unwrap());
+        let fft = fft.as_view();
+
+        let mut vec_buffers: Vec<_> = (0..NUM_CTS)
+            .map(|_| {
+                let mut buffers = ComputationBuffers::new();
+                buffers.resize(
+                    programmable_bootstrap_lwe_ciphertext_mem_optimized_requirement::<Scalar>(
+                        params.glwe_dimension.unwrap().to_glwe_size(),
+                        params.polynomial_size.unwrap(),
+                        fft,
+                    )
+                    .unwrap()
+                    .unaligned_bytes_required(),
+                );
+                buffers
+            })
+            .collect();
+
+        let glwe = GlweCiphertext::new(
+            Scalar::ONE << 60,
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            params.ciphertext_modulus.unwrap(),
+        );
+
+        let fbsk = FourierLweBootstrapKey::new(
+            params.lwe_dimension.unwrap(),
+            params.glwe_dimension.unwrap().to_glwe_size(),
+            params.polynomial_size.unwrap(),
+            params.pbs_base_log.unwrap(),
+            params.pbs_level.unwrap(),
+        );
+
+        for chunk_size in [1, 16, 32, 64, 128, 256, 512] {
+            let id = format!("{bench_name}_{name}_{chunk_size}chunk");
+            {
+                bench_group.bench_function(&id, |b| {
+                    b.iter(|| {
+                        lwe_vec
+                            .par_iter()
+                            .zip(output_lwe_list.par_iter_mut())
+                            .zip(vec_buffers.par_iter_mut())
+                            .take(chunk_size)
+                            .for_each(|((input_lwe, mut out_lwe), buffer)| {
+                                programmable_bootstrap_lwe_ciphertext_mem_optimized(
+                                    input_lwe,
+                                    &mut out_lwe,
+                                    &glwe,
+                                    &fbsk,
+                                    fft,
+                                    buffer.stack(),
+                                );
+                            });
+                        black_box(&mut output_lwe_list);
+                    })
+                });
+            }
+
+            let bit_size = (params.message_modulus.unwrap_or(2) as u32).ilog2();
+            write_to_json(
+                &id,
+                *params,
+                *name,
+                "pbs",
+                &OperatorType::Atomic,
+                bit_size,
+                vec![bit_size],
+            );
+        }
+    }
+}

tfhe/benches/integer/bench.rs

@@ -0,0 +1,977 @@
+#![allow(dead_code)]
+
+#[path = "../utilities.rs"]
+mod utilities;
+
+use crate::utilities::{write_to_json, OperatorType};
+use std::env;
+
+use criterion::{criterion_group, Criterion};
+use itertools::iproduct;
+use rand::rngs::ThreadRng;
+use rand::Rng;
+use std::vec::IntoIter;
+use tfhe::integer::keycache::KEY_CACHE;
+use tfhe::integer::{RadixCiphertext, ServerKey};
+use tfhe::shortint::keycache::NamedParam;
+
+#[allow(unused_imports)]
+use tfhe::shortint::parameters::{
+    PARAM_MESSAGE_1_CARRY_1_KS_PBS, PARAM_MESSAGE_2_CARRY_2_KS_PBS, PARAM_MESSAGE_3_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_4_KS_PBS, PARAM_MULTI_BIT_MESSAGE_2_CARRY_2_GROUP_2_KS_PBS,
+};
+
+/// An iterator that yields a succession of combinations
+/// of parameters and a num_block to achieve a certain bit_size ciphertext
+/// in radix decomposition
+struct ParamsAndNumBlocksIter {
+    params_and_bit_sizes:
+        itertools::Product<IntoIter<tfhe::shortint::PBSParameters>, IntoIter<usize>>,
+}
+
+impl Default for ParamsAndNumBlocksIter {
+    fn default() -> Self {
+        let is_multi_bit = match env::var("__TFHE_RS_BENCH_TYPE") {
+            Ok(val) => val.to_lowercase() == "multi_bit",
+            Err(_) => false,
+        };
+
+        if is_multi_bit {
+            let params = vec![PARAM_MULTI_BIT_MESSAGE_2_CARRY_2_GROUP_2_KS_PBS.into()];
+            let bit_sizes = vec![8, 16, 32, 40, 64];
+            let params_and_bit_sizes = iproduct!(params, bit_sizes);
+            Self {
+                params_and_bit_sizes,
+            }
+        } else {
+            // FIXME One set of parameter is tested since we want to benchmark only quickest
+            // operations.
+            let params = vec![
+                PARAM_MESSAGE_2_CARRY_2_KS_PBS.into(),
+                // PARAM_MESSAGE_3_CARRY_3_KS_PBS.into(),
+                // PARAM_MESSAGE_4_CARRY_4_KS_PBS.into(),
+            ];
+            let bit_sizes = vec![8, 16, 32, 40, 64, 128, 256];
+            let params_and_bit_sizes = iproduct!(params, bit_sizes);
+            Self {
+                params_and_bit_sizes,
+            }
+        }
+    }
+}
+
+impl Iterator for ParamsAndNumBlocksIter {
+    type Item = (tfhe::shortint::PBSParameters, usize, usize);
+
+    fn next(&mut self) -> Option<Self::Item> {
+        let (param, bit_size) = self.params_and_bit_sizes.next()?;
+        let num_block =
+            (bit_size as f64 / (param.message_modulus().0 as f64).log(2.0)).ceil() as usize;
+
+        Some((param, num_block, bit_size))
+    }
+}
+
+/// Base function to bench a server key function that is a binary operation, input ciphertexts will
+/// contain non zero carries
+fn bench_server_key_binary_function_dirty_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext, &mut RadixCiphertext),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_two_values = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let mut ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_1 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let mut ct_1 = cks.encrypt_radix(clear_1, num_block);
+
+                // Raise the degree, so as to ensure worst case path in operations
+                let mut carry_mod = param.carry_modulus().0;
+                while carry_mod > 0 {
+                    // Raise the degree, so as to ensure worst case path in operations
+                    let clearlow = rng.gen::<u128>();
+                    let clearhigh = rng.gen::<u128>();
+                    let clear_2 = tfhe::integer::U256::from((clearlow, clearhigh));
+                    let ct_2 = cks.encrypt_radix(clear_2, num_block);
+                    sks.unchecked_add_assign(&mut ct_0, &ct_2);
+                    sks.unchecked_add_assign(&mut ct_1, &ct_2);
+
+                    carry_mod -= 1;
+                }
+
+                (ct_0, ct_1)
+            };
+
+            b.iter_batched(
+                encrypt_two_values,
+                |(mut ct_0, mut ct_1)| {
+                    binary_op(&sks, &mut ct_0, &mut ct_1);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+/// Base function to bench a server key function that is a binary operation, input ciphertext will
+/// contain only zero carries
+fn bench_server_key_binary_function_clean_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext, &mut RadixCiphertext),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_two_values = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_1 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let ct_1 = cks.encrypt_radix(clear_1, num_block);
+
+                (ct_0, ct_1)
+            };
+
+            b.iter_batched(
+                encrypt_two_values,
+                |(mut ct_0, mut ct_1)| {
+                    binary_op(&sks, &mut ct_0, &mut ct_1);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+/// Base function to bench a server key function that is a unary operation, input ciphertexts will
+/// contain non zero carries
+fn bench_server_key_unary_function_dirty_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    unary_fn: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_one_value = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+
+                let mut ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                // Raise the degree, so as to ensure worst case path in operations
+                let mut carry_mod = param.carry_modulus().0;
+                while carry_mod > 0 {
+                    // Raise the degree, so as to ensure worst case path in operations
+                    let clearlow = rng.gen::<u128>();
+                    let clearhigh = rng.gen::<u128>();
+                    let clear_2 = tfhe::integer::U256::from((clearlow, clearhigh));
+                    let ct_2 = cks.encrypt_radix(clear_2, num_block);
+                    sks.unchecked_add_assign(&mut ct_0, &ct_2);
+
+                    carry_mod -= 1;
+                }
+
+                ct_0
+            };
+
+            b.iter_batched(
+                encrypt_one_value,
+                |mut ct_0| {
+                    unary_fn(&sks, &mut ct_0);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+/// Base function to bench a server key function that is a unary operation, input ciphertext will
+/// contain only zero carries
+fn bench_server_key_unary_function_clean_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    unary_fn: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_one_value = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+
+                cks.encrypt_radix(clear_0, num_block)
+            };
+
+            b.iter_batched(
+                encrypt_one_value,
+                |mut ct_0| {
+                    unary_fn(&sks, &mut ct_0);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+fn bench_server_key_binary_scalar_function_dirty_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext, u64),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_one_value = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let mut ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                // Raise the degree, so as to ensure worst case path in operations
+                let mut carry_mod = param.carry_modulus().0;
+                while carry_mod > 0 {
+                    // Raise the degree, so as to ensure worst case path in operations
+                    let clearlow = rng.gen::<u128>();
+                    let clearhigh = rng.gen::<u128>();
+                    let clear_2 = tfhe::integer::U256::from((clearlow, clearhigh));
+                    let ct_2 = cks.encrypt_radix(clear_2, num_block);
+                    sks.unchecked_add_assign(&mut ct_0, &ct_2);
+
+                    carry_mod -= 1;
+                }
+
+                let clear_1 = rng.gen::<u64>();
+
+                (ct_0, clear_1)
+            };
+
+            b.iter_batched(
+                encrypt_one_value,
+                |(mut ct_0, clear_1)| {
+                    binary_op(&sks, &mut ct_0, clear_1);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+fn bench_server_key_binary_scalar_function_clean_inputs<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+) where
+    F: Fn(&ServerKey, &mut RadixCiphertext, u64),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_one_value = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                let clear_1 = rng.gen::<u64>();
+
+                (ct_0, clear_1)
+            };
+
+            b.iter_batched(
+                encrypt_one_value,
+                |(mut ct_0, clear_1)| {
+                    binary_op(&sks, &mut ct_0, clear_1);
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+// Functions used to apply different way of selecting a scalar based on the context.
+fn default_scalar(rng: &mut ThreadRng, _clear_bit_size: usize) -> u64 {
+    rng.gen::<u64>()
+}
+
+fn shift_scalar(_rng: &mut ThreadRng, _clear_bit_size: usize) -> u64 {
+    // Shifting by one is the worst case scenario.
+    1
+}
+
+fn mul_scalar(rng: &mut ThreadRng, _clear_bit_size: usize) -> u64 {
+    loop {
+        let scalar = rng.gen_range(3u64..=u64::MAX);
+        // If scalar is power of two, it is just a shit, which is an happy path.
+        if !scalar.is_power_of_two() {
+            return scalar;
+        }
+    }
+}
+
+fn div_scalar(rng: &mut ThreadRng, clear_bit_size: usize) -> u64 {
+    loop {
+        let scalar = rng.gen_range(1..=u64::MAX);
+        // Avoid overflow issues for u64 where we would take values mod 1
+        if (scalar as u128 % (1u128 << clear_bit_size)) != 0 {
+            return scalar;
+        }
+    }
+}
+
+fn if_then_else_parallelized(c: &mut Criterion) {
+    let bench_name = "integer::if_then_else_parallelized";
+    let display_name = "if_then_else";
+
+    let mut bench_group = c.benchmark_group(bench_name);
+    bench_group
+        .sample_size(15)
+        .measurement_time(std::time::Duration::from_secs(60));
+    let mut rng = rand::thread_rng();
+
+    for (param, num_block, bit_size) in ParamsAndNumBlocksIter::default() {
+        let param_name = param.name();
+
+        let bench_id = format!("{bench_name}::{param_name}::{bit_size}_bits");
+        bench_group.bench_function(&bench_id, |b| {
+            let (cks, sks) = KEY_CACHE.get_from_params(param);
+
+            let encrypt_tree_values = || {
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_0 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let ct_0 = cks.encrypt_radix(clear_0, num_block);
+
+                let clearlow = rng.gen::<u128>();
+                let clearhigh = rng.gen::<u128>();
+                let clear_1 = tfhe::integer::U256::from((clearlow, clearhigh));
+                let ct_1 = cks.encrypt_radix(clear_1, num_block);
+
+                let cond = sks.create_trivial_radix(rng.gen_bool(0.5) as u64, num_block);
+
+                (cond, ct_0, ct_1)
+            };
+
+            b.iter_batched(
+                encrypt_tree_values,
+                |(condition, true_ct, false_ct)| {
+                    sks.if_then_else_parallelized(&condition, &true_ct, &false_ct)
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            bit_size as u32,
+            vec![param.message_modulus().0.ilog2(); num_block],
+        );
+    }
+
+    bench_group.finish()
+}
+
+macro_rules! define_server_key_bench_unary_fn (
+    (method_name: $server_key_method:ident, display_name:$name:ident) => {
+        fn $server_key_method(c: &mut Criterion) {
+            bench_server_key_unary_function_dirty_inputs(
+                c,
+                concat!("ServerKey::", stringify!($server_key_method)),
+                stringify!($name),
+                |server_key, lhs| {
+                  server_key.$server_key_method(lhs);
+            })
+        }
+    }
+);
+
+macro_rules! define_server_key_bench_unary_default_fn (
+    (method_name: $server_key_method:ident, display_name:$name:ident) => {
+        fn $server_key_method(c: &mut Criterion) {
+            bench_server_key_unary_function_clean_inputs(
+                c,
+                concat!("ServerKey::", stringify!($server_key_method)),
+                stringify!($name),
+                |server_key, lhs| {
+                  server_key.$server_key_method(lhs);
+            })
+        }
+    }
+);
+
+macro_rules! define_server_key_bench_fn (
+  (method_name: $server_key_method:ident, display_name:$name:ident) => {
+      fn $server_key_method(c: &mut Criterion) {
+        bench_server_key_binary_function_dirty_inputs(
+              c,
+              concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
+              |server_key, lhs, rhs| {
+                server_key.$server_key_method(lhs, rhs);
+          })
+      }
+  }
+);
+
+macro_rules! define_server_key_bench_default_fn (
+    (method_name: $server_key_method:ident, display_name:$name:ident) => {
+        fn $server_key_method(c: &mut Criterion) {
+          bench_server_key_binary_function_clean_inputs(
+                c,
+                concat!("ServerKey::", stringify!($server_key_method)),
+                stringify!($name),
+                |server_key, lhs, rhs| {
+                  server_key.$server_key_method(lhs, rhs);
+            })
+        }
+    }
+  );
+
+macro_rules! define_server_key_bench_scalar_fn (
+  (method_name: $server_key_method:ident, display_name:$name:ident) => {
+      fn $server_key_method(c: &mut Criterion) {
+          bench_server_key_binary_scalar_function_dirty_inputs(
+              c,
+              concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
+              |server_key, lhs, rhs| {
+                server_key.$server_key_method(lhs, rhs);
+          })
+      }
+  }
+);
+
+macro_rules! define_server_key_bench_scalar_default_fn (
+    (method_name: $server_key_method:ident, display_name:$name:ident) => {
+        fn $server_key_method(c: &mut Criterion) {
+            bench_server_key_binary_scalar_function_clean_inputs(
+                c,
+                concat!("ServerKey::", stringify!($server_key_method)),
+                stringify!($name),
+                |server_key, lhs, rhs| {
+                  server_key.$server_key_method(lhs, rhs);
+            })
+        }
+    }
+  );
+
+define_server_key_bench_fn!(method_name: smart_add, display_name: add);
+define_server_key_bench_fn!(method_name: smart_sub, display_name: sub);
+define_server_key_bench_fn!(method_name: smart_mul, display_name: mul);
+define_server_key_bench_fn!(method_name: smart_bitand, display_name: bitand);
+define_server_key_bench_fn!(method_name: smart_bitor, display_name: bitor);
+define_server_key_bench_fn!(method_name: smart_bitxor, display_name: bitxor);
+
+define_server_key_bench_fn!(method_name: smart_add_parallelized, display_name: add);
+define_server_key_bench_fn!(method_name: smart_sub_parallelized, display_name: sub);
+define_server_key_bench_fn!(method_name: smart_mul_parallelized, display_name: mul);
+define_server_key_bench_fn!(method_name: smart_bitand_parallelized, display_name: bitand);
+define_server_key_bench_fn!(method_name: smart_bitxor_parallelized, display_name: bitxor);
+define_server_key_bench_fn!(method_name: smart_bitor_parallelized, display_name: bitor);
+
+define_server_key_bench_default_fn!(method_name: add_parallelized, display_name: add);
+define_server_key_bench_default_fn!(method_name: sub_parallelized, display_name: sub);
+define_server_key_bench_default_fn!(method_name: mul_parallelized, display_name: mul);
+define_server_key_bench_default_fn!(method_name: bitand_parallelized, display_name: bitand);
+define_server_key_bench_default_fn!(method_name: bitxor_parallelized, display_name: bitxor);
+define_server_key_bench_default_fn!(method_name: bitor_parallelized, display_name: bitor);
+define_server_key_bench_unary_default_fn!(method_name: bitnot_parallelized, display_name: bitnot);
+
+define_server_key_bench_fn!(method_name: unchecked_add, display_name: add);
+define_server_key_bench_fn!(method_name: unchecked_sub, display_name: sub);
+define_server_key_bench_fn!(method_name: unchecked_mul, display_name: mul);
+define_server_key_bench_fn!(method_name: unchecked_bitand, display_name: bitand);
+define_server_key_bench_fn!(method_name: unchecked_bitor, display_name: bitor);
+define_server_key_bench_fn!(method_name: unchecked_bitxor, display_name: bitxor);
+
+define_server_key_bench_fn!(method_name: unchecked_mul_parallelized, display_name: mul);
+define_server_key_bench_fn!(
+    method_name: unchecked_bitand_parallelized,
+    display_name: bitand
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_bitor_parallelized,
+    display_name: bitor
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_bitxor_parallelized,
+    display_name: bitxor
+);
+
+define_server_key_bench_scalar_fn!(method_name: smart_scalar_add, display_name: add);
+define_server_key_bench_scalar_fn!(method_name: smart_scalar_sub, display_name: sub);
+define_server_key_bench_scalar_fn!(method_name: smart_scalar_mul, display_name: mul);
+
+define_server_key_bench_scalar_fn!(
+    method_name: smart_scalar_add_parallelized,
+    display_name: add
+);
+define_server_key_bench_scalar_fn!(
+    method_name: smart_scalar_sub_parallelized,
+    display_name: sub
+);
+define_server_key_bench_scalar_fn!(
+    method_name: smart_scalar_mul_parallelized,
+    display_name: mul
+);
+
+define_server_key_bench_scalar_default_fn!(method_name: scalar_add_parallelized, display_name: add);
+define_server_key_bench_scalar_default_fn!(method_name: scalar_sub_parallelized, display_name: sub);
+define_server_key_bench_scalar_default_fn!(method_name: scalar_mul_parallelized, display_name: mul);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_left_shift_parallelized,
+    display_name: left_shift
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_right_shift_parallelized,
+    display_name: right_shift
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_eq_parallelized,
+    display_name: scalar_equal
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_ne_parallelized,
+    display_name: scalar_not_equal
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_le_parallelized,
+    display_name: scalar_less_or_equal
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_lt_parallelized,
+    display_name: scalar_less_than
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_ge_parallelized,
+    display_name: scalar_greater_or_equal
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_gt_parallelized,
+    display_name: scalar_greater_than
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_max_parallelized,
+    display_name: scalar_max
+);
+define_server_key_bench_scalar_default_fn!(
+    method_name: scalar_min_parallelized,
+    display_name: scalar_min
+);
+
+define_server_key_bench_scalar_fn!(method_name: unchecked_scalar_add, display_name: add);
+define_server_key_bench_scalar_fn!(method_name: unchecked_scalar_sub, display_name: sub);
+define_server_key_bench_scalar_fn!(method_name: unchecked_small_scalar_mul, display_name: mul);
+
+define_server_key_bench_unary_fn!(method_name: smart_neg, display_name: negation);
+define_server_key_bench_unary_fn!(method_name: smart_neg_parallelized, display_name: negation);
+define_server_key_bench_unary_default_fn!(method_name: neg_parallelized, display_name: negation);
+
+define_server_key_bench_unary_fn!(method_name: full_propagate, display_name: carry_propagation);
+define_server_key_bench_unary_fn!(
+    method_name: full_propagate_parallelized,
+    display_name: carry_propagation
+);
+
+define_server_key_bench_fn!(method_name: unchecked_max, display_name: max);
+define_server_key_bench_fn!(method_name: unchecked_min, display_name: min);
+define_server_key_bench_fn!(method_name: unchecked_eq, display_name: equal);
+define_server_key_bench_fn!(method_name: unchecked_lt, display_name: less_than);
+define_server_key_bench_fn!(method_name: unchecked_le, display_name: less_or_equal);
+define_server_key_bench_fn!(method_name: unchecked_gt, display_name: greater_than);
+define_server_key_bench_fn!(method_name: unchecked_ge, display_name: greater_or_equal);
+
+define_server_key_bench_fn!(method_name: unchecked_max_parallelized, display_name: max);
+define_server_key_bench_fn!(method_name: unchecked_min_parallelized, display_name: min);
+define_server_key_bench_fn!(method_name: unchecked_eq_parallelized, display_name: equal);
+define_server_key_bench_fn!(
+    method_name: unchecked_lt_parallelized,
+    display_name: less_than
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_le_parallelized,
+    display_name: less_or_equal
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_gt_parallelized,
+    display_name: greater_than
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_ge_parallelized,
+    display_name: greater_or_equal
+);
+
+define_server_key_bench_fn!(method_name: smart_max, display_name: max);
+define_server_key_bench_fn!(method_name: smart_min, display_name: min);
+define_server_key_bench_fn!(method_name: smart_eq, display_name: equal);
+define_server_key_bench_fn!(method_name: smart_lt, display_name: less_than);
+define_server_key_bench_fn!(method_name: smart_le, display_name: less_or_equal);
+define_server_key_bench_fn!(method_name: smart_gt, display_name: greater_than);
+define_server_key_bench_fn!(method_name: smart_ge, display_name: greater_or_equal);
+
+define_server_key_bench_fn!(method_name: smart_max_parallelized, display_name: max);
+define_server_key_bench_fn!(method_name: smart_min_parallelized, display_name: min);
+define_server_key_bench_fn!(method_name: smart_eq_parallelized, display_name: equal);
+define_server_key_bench_fn!(method_name: smart_lt_parallelized, display_name: less_than);
+define_server_key_bench_fn!(
+    method_name: smart_le_parallelized,
+    display_name: less_or_equal
+);
+define_server_key_bench_fn!(
+    method_name: smart_gt_parallelized,
+    display_name: greater_than
+);
+define_server_key_bench_fn!(
+    method_name: smart_ge_parallelized,
+    display_name: greater_or_equal
+);
+
+define_server_key_bench_default_fn!(method_name: max_parallelized, display_name: max);
+define_server_key_bench_default_fn!(method_name: min_parallelized, display_name: min);
+define_server_key_bench_default_fn!(method_name: eq_parallelized, display_name: equal);
+define_server_key_bench_default_fn!(method_name: ne_parallelized, display_name: not_equal);
+define_server_key_bench_default_fn!(method_name: lt_parallelized, display_name: less_than);
+define_server_key_bench_default_fn!(method_name: le_parallelized, display_name: less_or_equal);
+define_server_key_bench_default_fn!(method_name: gt_parallelized, display_name: greater_than);
+define_server_key_bench_default_fn!(method_name: ge_parallelized, display_name: greater_or_equal);
+
+define_server_key_bench_default_fn!(
+    method_name: left_shift_parallelized,
+    display_name: left_shift
+);
+define_server_key_bench_default_fn!(
+    method_name: right_shift_parallelized,
+    display_name: right_shift
+);
+define_server_key_bench_default_fn!(
+    method_name: rotate_left_parallelized,
+    display_name: rotate_left
+);
+define_server_key_bench_default_fn!(
+    method_name: rotate_right_parallelized,
+    display_name: rotate_right
+);
+
+criterion_group!(
+    smart_ops,
+    smart_neg,
+    smart_add,
+    smart_mul,
+    smart_bitand,
+    smart_bitor,
+    smart_bitxor,
+    smart_max,
+    smart_min,
+    smart_eq,
+    smart_lt,
+    smart_le,
+    smart_gt,
+    smart_ge,
+);
+
+criterion_group!(
+    smart_parallelized_ops,
+    smart_add_parallelized,
+    smart_sub_parallelized,
+    smart_mul_parallelized,
+    smart_bitand_parallelized,
+    smart_bitor_parallelized,
+    smart_bitxor_parallelized,
+    smart_max_parallelized,
+    smart_min_parallelized,
+    smart_eq_parallelized,
+    smart_lt_parallelized,
+    smart_le_parallelized,
+    smart_gt_parallelized,
+    smart_ge_parallelized,
+);
+
+criterion_group!(
+    default_parallelized_ops,
+    add_parallelized,
+    sub_parallelized,
+    mul_parallelized,
+    neg_parallelized,
+    bitand_parallelized,
+    bitnot_parallelized,
+    bitor_parallelized,
+    bitxor_parallelized,
+    max_parallelized,
+    min_parallelized,
+    eq_parallelized,
+    ne_parallelized,
+    lt_parallelized,
+    le_parallelized,
+    gt_parallelized,
+    ge_parallelized,
+    left_shift_parallelized,
+    right_shift_parallelized,
+    rotate_left_parallelized,
+    rotate_right_parallelized,
+    if_then_else_parallelized,
+);
+
+criterion_group!(
+    smart_scalar_ops,
+    smart_scalar_add,
+    smart_scalar_sub,
+    smart_scalar_mul,
+);
+
+criterion_group!(
+    smart_scalar_parallelized_ops,
+    smart_scalar_add_parallelized,
+    smart_scalar_sub_parallelized,
+    smart_scalar_mul_parallelized,
+);
+
+criterion_group!(
+    default_scalar_parallelized_ops,
+    scalar_add_parallelized,
+    scalar_sub_parallelized,
+    scalar_mul_parallelized,
+    scalar_left_shift_parallelized,
+    scalar_right_shift_parallelized,
+    scalar_eq_parallelized,
+    scalar_ne_parallelized,
+    scalar_lt_parallelized,
+    scalar_le_parallelized,
+    scalar_gt_parallelized,
+    scalar_ge_parallelized,
+    scalar_min_parallelized,
+    scalar_max_parallelized,
+);
+
+criterion_group!(
+    unchecked_ops,
+    unchecked_add,
+    unchecked_sub,
+    unchecked_mul,
+    unchecked_bitand,
+    unchecked_bitor,
+    unchecked_bitxor,
+    unchecked_max,
+    unchecked_min,
+    unchecked_eq,
+    unchecked_lt,
+    unchecked_le,
+    unchecked_gt,
+    unchecked_ge,
+);
+
+criterion_group!(
+    unchecked_scalar_ops,
+    unchecked_scalar_add,
+    unchecked_scalar_sub,
+    unchecked_small_scalar_mul,
+    unchecked_max_parallelized,
+    unchecked_min_parallelized,
+    unchecked_eq_parallelized,
+    unchecked_lt_parallelized,
+    unchecked_le_parallelized,
+    unchecked_gt_parallelized,
+    unchecked_ge_parallelized,
+    unchecked_bitand_parallelized,
+    unchecked_bitor_parallelized,
+    unchecked_bitxor_parallelized,
+);
+
+criterion_group!(misc, full_propagate, full_propagate_parallelized);
+
+fn main() {
+    match env::var("__TFHE_RS_BENCH_OP_FLAVOR") {
+        Ok(val) => {
+            match val.to_lowercase().as_str() {
+                "default" => default_parallelized_ops(),
+                "default_scalar" => default_scalar_parallelized_ops(),
+                "smart" => smart_ops(),
+                "smart_scalar" => smart_scalar_ops(),
+                "smart_parallelized" => smart_parallelized_ops(),
+                "smart_scalar_parallelized" => smart_scalar_parallelized_ops(),
+                "unchecked" => unchecked_ops(),
+                "unchecked_scalar" => unchecked_scalar_ops(),
+                "misc" => misc(),
+                _ => panic!("unknown benchmark operations flavor"),
+            };
+        }
+        Err(_) => {
+            default_parallelized_ops();
+            default_scalar_parallelized_ops()
+        }
+    };
+
+    Criterion::default().configure_from_args().final_summary();
+}

tfhe/benches/keygen/bench.rs

@@ -0,0 +1,45 @@
+use concrete_csprng::seeders::Seeder;
+use criterion::*;
+use tfhe::core_crypto::commons::generators::DeterministicSeeder;
+use tfhe::core_crypto::prelude::{
+    allocate_and_generate_new_binary_glwe_secret_key,
+    par_allocate_and_generate_new_lwe_bootstrap_key, ActivatedRandomGenerator, CiphertextModulus,
+    EncryptionRandomGenerator, SecretRandomGenerator,
+};
+use tfhe::core_crypto::seeders::new_seeder;
+use tfhe::shortint::prelude::*;
+
+fn criterion_bench(c: &mut Criterion) {
+    let parameters = PARAM_MESSAGE_2_CARRY_2_KS_PBS;
+    let mut seeder = new_seeder();
+    let mut deterministic_seeder =
+        DeterministicSeeder::<ActivatedRandomGenerator>::new(seeder.seed());
+    let mut secret_generator =
+        SecretRandomGenerator::<ActivatedRandomGenerator>::new(deterministic_seeder.seed());
+    let mut encryption_generator = EncryptionRandomGenerator::<ActivatedRandomGenerator>::new(
+        deterministic_seeder.seed(),
+        &mut deterministic_seeder,
+    );
+    let glwe_secret_key = allocate_and_generate_new_binary_glwe_secret_key::<u64, _>(
+        parameters.glwe_dimension,
+        parameters.polynomial_size,
+        &mut secret_generator,
+    );
+    let lwe_secret_key_after_ks = glwe_secret_key.clone().into_lwe_secret_key();
+    c.bench_function("keygen", |b| {
+        b.iter(|| {
+            let _ = par_allocate_and_generate_new_lwe_bootstrap_key(
+                &lwe_secret_key_after_ks,
+                &glwe_secret_key,
+                parameters.pbs_base_log,
+                parameters.pbs_level,
+                parameters.glwe_modular_std_dev,
+                CiphertextModulus::new_native(),
+                &mut encryption_generator,
+            );
+        });
+    });
+}
+
+criterion_group!(benches, criterion_bench);
+criterion_main!(benches);

tfhe/benches/shortint/bench.rs

@@ -1,39 +1,109 @@
-use criterion::{criterion_group, criterion_main, Criterion};
+#[path = "../utilities.rs"]
+mod utilities;
+
+use crate::utilities::{write_to_json, OperatorType};
+use std::env;
+
+use criterion::{criterion_group, Criterion};
+use tfhe::shortint::keycache::NamedParam;
 use tfhe::shortint::parameters::*;
-use tfhe::shortint::{Ciphertext, Parameters, ServerKey};
+use tfhe::shortint::{Ciphertext, ClassicPBSParameters, ServerKey, ShortintParameterSet};
 
 use rand::Rng;
 use tfhe::shortint::keycache::KEY_CACHE;
 
 use tfhe::shortint::keycache::KEY_CACHE_WOPBS;
-use tfhe::shortint::parameters::parameters_wopbs::WOPBS_PARAM_MESSAGE_4_NORM2_6;
+use tfhe::shortint::parameters::parameters_wopbs::WOPBS_PARAM_MESSAGE_4_NORM2_6_KS_PBS;
 
-macro_rules! named_param {
-    ($param:ident) => {
-        (stringify!($param), $param)
-    };
-}
-
-const SERVER_KEY_BENCH_PARAMS: [(&str, Parameters); 4] = [
-    named_param!(PARAM_MESSAGE_1_CARRY_1),
-    named_param!(PARAM_MESSAGE_2_CARRY_2),
-    named_param!(PARAM_MESSAGE_3_CARRY_3),
-    named_param!(PARAM_MESSAGE_4_CARRY_4),
+const SERVER_KEY_BENCH_PARAMS: [ClassicPBSParameters; 4] = [
+    PARAM_MESSAGE_1_CARRY_1_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_2_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_4_KS_PBS,
 ];
 
-fn bench_server_key_binary_function<F>(c: &mut Criterion, bench_name: &str, binary_op: F)
-where
-    F: Fn(&ServerKey, &mut Ciphertext, &mut Ciphertext),
+const SERVER_KEY_BENCH_PARAMS_EXTENDED: [ClassicPBSParameters; 15] = [
+    PARAM_MESSAGE_1_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_1_CARRY_1_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_1_KS_PBS,
+    PARAM_MESSAGE_2_CARRY_2_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_2_KS_PBS,
+    PARAM_MESSAGE_3_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_3_KS_PBS,
+    PARAM_MESSAGE_4_CARRY_4_KS_PBS,
+    PARAM_MESSAGE_5_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_6_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_7_CARRY_0_KS_PBS,
+    PARAM_MESSAGE_8_CARRY_0_KS_PBS,
+];
+
+fn bench_server_key_unary_function<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    unary_op: F,
+    params: &[ClassicPBSParameters],
+) where
+    F: Fn(&ServerKey, &mut Ciphertext),
 {
     let mut bench_group = c.benchmark_group(bench_name);
 
-    for (param_name, param) in SERVER_KEY_BENCH_PARAMS {
+    for param in params.iter() {
+        let param: PBSParameters = (*param).into();
         let keys = KEY_CACHE.get_from_param(param);
         let (cks, sks) = (keys.client_key(), keys.server_key());
 
         let mut rng = rand::thread_rng();
 
-        let modulus = 1_u64 << cks.parameters.message_modulus.0;
+        let modulus = cks.parameters.message_modulus().0 as u64;
+
+        let clear_text = rng.gen::<u64>() % modulus;
+
+        let mut ct = cks.encrypt(clear_text);
+
+        let bench_id = format!("{bench_name}::{}", param.name());
+        bench_group.bench_function(&bench_id, |b| {
+            b.iter(|| {
+                unary_op(sks, &mut ct);
+            })
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
+    }
+
+    bench_group.finish()
+}
+
+fn bench_server_key_binary_function<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+    params: &[ClassicPBSParameters],
+) where
+    F: Fn(&ServerKey, &mut Ciphertext, &mut Ciphertext),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    for param in params.iter() {
+        let param: PBSParameters = (*param).into();
+        let keys = KEY_CACHE.get_from_param(param);
+        let (cks, sks) = (keys.client_key(), keys.server_key());
+
+        let mut rng = rand::thread_rng();
+
+        let modulus = cks.parameters.message_modulus().0 as u64;
 
         let clear_0 = rng.gen::<u64>() % modulus;
         let clear_1 = rng.gen::<u64>() % modulus;
@@ -41,42 +111,118 @@ where
         let mut ct_0 = cks.encrypt(clear_0);
         let mut ct_1 = cks.encrypt(clear_1);
 
-        let bench_id = format!("{bench_name}::{param_name}");
+        let bench_id = format!("{bench_name}::{}", param.name());
         bench_group.bench_function(&bench_id, |b| {
             b.iter(|| {
                 binary_op(sks, &mut ct_0, &mut ct_1);
             })
         });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
     }
 
     bench_group.finish()
 }
 
-fn bench_server_key_binary_scalar_function<F>(c: &mut Criterion, bench_name: &str, binary_op: F)
-where
+fn bench_server_key_binary_scalar_function<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+    params: &[ClassicPBSParameters],
+) where
     F: Fn(&ServerKey, &mut Ciphertext, u8),
 {
     let mut bench_group = c.benchmark_group(bench_name);
 
-    for (param_name, param) in SERVER_KEY_BENCH_PARAMS {
+    for param in params {
+        let param: PBSParameters = (*param).into();
         let keys = KEY_CACHE.get_from_param(param);
         let (cks, sks) = (keys.client_key(), keys.server_key());
 
         let mut rng = rand::thread_rng();
 
-        let modulus = 1_u64 << cks.parameters.message_modulus.0;
+        let modulus = cks.parameters.message_modulus().0 as u64;
 
         let clear_0 = rng.gen::<u64>() % modulus;
         let clear_1 = rng.gen::<u64>() % modulus;
 
         let mut ct_0 = cks.encrypt(clear_0);
 
-        let bench_id = format!("{bench_name}::{param_name}");
+        let bench_id = format!("{bench_name}::{}", param.name());
         bench_group.bench_function(&bench_id, |b| {
             b.iter(|| {
                 binary_op(sks, &mut ct_0, clear_1 as u8);
             })
         });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
+    }
+
+    bench_group.finish()
+}
+
+fn bench_server_key_binary_scalar_division_function<F>(
+    c: &mut Criterion,
+    bench_name: &str,
+    display_name: &str,
+    binary_op: F,
+    params: &[ClassicPBSParameters],
+) where
+    F: Fn(&ServerKey, &mut Ciphertext, u8),
+{
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    for param in params {
+        let param: PBSParameters = (*param).into();
+        let keys = KEY_CACHE.get_from_param(param);
+        let (cks, sks) = (keys.client_key(), keys.server_key());
+
+        let mut rng = rand::thread_rng();
+
+        let modulus = cks.parameters.message_modulus().0 as u64;
+        assert_ne!(modulus, 1);
+
+        let clear_0 = rng.gen::<u64>() % modulus;
+        let mut clear_1 = rng.gen::<u64>() % modulus;
+        while clear_1 == 0 {
+            clear_1 = rng.gen::<u64>() % modulus;
+        }
+
+        let mut ct_0 = cks.encrypt(clear_0);
+
+        let bench_id = format!("{bench_name}::{}", param.name());
+        bench_group.bench_function(&bench_id, |b| {
+            b.iter(|| {
+                binary_op(sks, &mut ct_0, clear_1 as u8);
+            })
+        });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            display_name,
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
     }
 
     bench_group.finish()
@@ -85,24 +231,35 @@ where
 fn carry_extract(c: &mut Criterion) {
     let mut bench_group = c.benchmark_group("carry_extract");
 
-    for (param_name, param) in SERVER_KEY_BENCH_PARAMS {
+    for param in SERVER_KEY_BENCH_PARAMS {
+        let param: PBSParameters = param.into();
         let keys = KEY_CACHE.get_from_param(param);
         let (cks, sks) = (keys.client_key(), keys.server_key());
 
         let mut rng = rand::thread_rng();
 
-        let modulus = 1_u64 << cks.parameters.message_modulus.0;
+        let modulus = cks.parameters.message_modulus().0 as u64;
 
         let clear_0 = rng.gen::<u64>() % modulus;
 
         let ct_0 = cks.encrypt(clear_0);
 
-        let bench_id = format!("ServerKey::carry_extract::{param_name}");
+        let bench_id = format!("ServerKey::carry_extract::{}", param.name());
         bench_group.bench_function(&bench_id, |b| {
             b.iter(|| {
-                sks.carry_extract(&ct_0);
+                let _ = sks.carry_extract(&ct_0);
             })
         });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            "carry_extract",
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
     }
 
     bench_group.finish()
@@ -111,38 +268,52 @@ fn carry_extract(c: &mut Criterion) {
 fn programmable_bootstrapping(c: &mut Criterion) {
     let mut bench_group = c.benchmark_group("programmable_bootstrap");
 
-    for (param_name, param) in SERVER_KEY_BENCH_PARAMS {
+    for param in SERVER_KEY_BENCH_PARAMS {
+        let param: PBSParameters = param.into();
         let keys = KEY_CACHE.get_from_param(param);
         let (cks, sks) = (keys.client_key(), keys.server_key());
 
         let mut rng = rand::thread_rng();
 
-        let modulus = cks.parameters.message_modulus.0 as u64;
+        let modulus = cks.parameters.message_modulus().0 as u64;
 
-        let acc = sks.generate_accumulator(|x| x);
+        let acc = sks.generate_lookup_table(|x| x);
 
         let clear_0 = rng.gen::<u64>() % modulus;
 
         let ctxt = cks.encrypt(clear_0);
 
-        let id = format!("ServerKey::programmable_bootstrap::{param_name}");
+        let bench_id = format!("ServerKey::programmable_bootstrap::{}", param.name());
 
-        bench_group.bench_function(&id, |b| {
+        bench_group.bench_function(&bench_id, |b| {
             b.iter(|| {
-                sks.keyswitch_programmable_bootstrap(&ctxt, &acc);
+                let _ = sks.apply_lookup_table(&ctxt, &acc);
             })
         });
+
+        write_to_json::<u64, _>(
+            &bench_id,
+            param,
+            param.name(),
+            "pbs",
+            &OperatorType::Atomic,
+            param.message_modulus().0.ilog2(),
+            vec![param.message_modulus().0.ilog2()],
+        );
     }
 
     bench_group.finish();
 }
 
-fn bench_wopbs_param_message_8_norm2_5(c: &mut Criterion) {
+// TODO: remove?
+fn _bench_wopbs_param_message_8_norm2_5(c: &mut Criterion) {
     let mut bench_group = c.benchmark_group("programmable_bootstrap");
 
-    let param = WOPBS_PARAM_MESSAGE_4_NORM2_6;
+    let param = WOPBS_PARAM_MESSAGE_4_NORM2_6_KS_PBS;
+    let param_set: ShortintParameterSet = param.try_into().unwrap();
+    let pbs_params = param_set.pbs_parameters().unwrap();
 
-    let keys = KEY_CACHE_WOPBS.get_from_param((param, param));
+    let keys = KEY_CACHE_WOPBS.get_from_param((pbs_params, param));
     let (cks, _, wopbs_key) = (keys.client_key(), keys.server_key(), keys.wopbs_key());
 
     let mut rng = rand::thread_rng();
@@ -155,78 +326,419 @@ fn bench_wopbs_param_message_8_norm2_5(c: &mut Criterion) {
 
     bench_group.bench_function(&id, |b| {
         b.iter(|| {
-            wopbs_key.programmable_bootstrapping_native_crt(&mut ct, &vec_lut);
+            let _ = wopbs_key.programmable_bootstrapping_native_crt(&mut ct, &vec_lut);
         })
     });
 
     bench_group.finish();
 }
 
+macro_rules! define_server_key_unary_bench_fn (
+  (method_name:$server_key_method:ident, display_name:$name:ident, $params:expr) => {
+      fn $server_key_method(c: &mut Criterion) {
+          bench_server_key_unary_function(
+              c,
+              concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
+              |server_key, lhs| {
+                let _ = server_key.$server_key_method(lhs);},
+              $params)
+      }
+  }
+);
+
 macro_rules! define_server_key_bench_fn (
-  ($server_key_method:ident) => {
+  (method_name:$server_key_method:ident, display_name:$name:ident, $params:expr) => {
       fn $server_key_method(c: &mut Criterion) {
           bench_server_key_binary_function(
               c,
               concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
               |server_key, lhs, rhs| {
-                server_key.$server_key_method(lhs, rhs);
-          })
+                let _ = server_key.$server_key_method(lhs, rhs);},
+              $params)
       }
   }
 );
 
 macro_rules! define_server_key_scalar_bench_fn (
-  ($server_key_method:ident) => {
+  (method_name:$server_key_method:ident, display_name:$name:ident, $params:expr) => {
       fn $server_key_method(c: &mut Criterion) {
           bench_server_key_binary_scalar_function(
               c,
               concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
               |server_key, lhs, rhs| {
-                server_key.$server_key_method(lhs, rhs);
-          })
+                let _ = server_key.$server_key_method(lhs, rhs);},
+              $params)
       }
   }
 );
 
-define_server_key_bench_fn!(unchecked_add);
-define_server_key_bench_fn!(unchecked_sub);
-define_server_key_bench_fn!(unchecked_mul_lsb);
-define_server_key_bench_fn!(unchecked_mul_msb);
-define_server_key_bench_fn!(smart_bitand);
-define_server_key_bench_fn!(smart_bitor);
-define_server_key_bench_fn!(smart_bitxor);
-define_server_key_bench_fn!(smart_add);
-define_server_key_bench_fn!(smart_sub);
-define_server_key_bench_fn!(smart_mul_lsb);
+macro_rules! define_server_key_scalar_div_bench_fn (
+  (method_name:$server_key_method:ident, display_name:$name:ident, $params:expr) => {
+      fn $server_key_method(c: &mut Criterion) {
+          bench_server_key_binary_scalar_division_function(
+              c,
+              concat!("ServerKey::", stringify!($server_key_method)),
+              stringify!($name),
+              |server_key, lhs, rhs| {
+                let _ = server_key.$server_key_method(lhs, rhs);},
+              $params)
+      }
+  }
+);
 
-define_server_key_scalar_bench_fn!(unchecked_scalar_add);
-define_server_key_scalar_bench_fn!(unchecked_scalar_mul);
+define_server_key_unary_bench_fn!(
+    method_name: unchecked_neg,
+    display_name: negation,
+    &SERVER_KEY_BENCH_PARAMS
+);
+
+define_server_key_bench_fn!(
+    method_name: unchecked_add,
+    display_name: add,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_sub,
+    display_name: sub,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_mul_lsb,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_mul_msb,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_div,
+    display_name: div,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_bench_fn!(
+    method_name: smart_bitand,
+    display_name: bitand,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: smart_bitor,
+    display_name: bitor,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: smart_bitxor,
+    display_name: bitxor,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: smart_add,
+    display_name: add,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: smart_sub,
+    display_name: sub,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: smart_mul_lsb,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: bitand,
+    display_name: bitand,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: bitor,
+    display_name: bitor,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: bitxor,
+    display_name: bitxor,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: add,
+    display_name: add,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: sub,
+    display_name: sub,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: mul,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: div,
+    display_name: div,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: greater,
+    display_name: greater,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: greater_or_equal,
+    display_name: greater_or_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: less,
+    display_name: less,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: less_or_equal,
+    display_name: less_or_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: equal,
+    display_name: equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: not_equal,
+    display_name: not_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_unary_bench_fn!(
+    method_name: neg,
+    display_name: negation,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_greater,
+    display_name: greater_than,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_less,
+    display_name: less_than,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_bench_fn!(
+    method_name: unchecked_equal,
+    display_name: equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+
+define_server_key_scalar_bench_fn!(
+    method_name: unchecked_scalar_add,
+    display_name: add,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_scalar_bench_fn!(
+    method_name: unchecked_scalar_sub,
+    display_name: sub,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_scalar_bench_fn!(
+    method_name: unchecked_scalar_mul,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_scalar_bench_fn!(
+    method_name: unchecked_scalar_left_shift,
+    display_name: left_shift,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: unchecked_scalar_right_shift,
+    display_name: right_shift,
+    &SERVER_KEY_BENCH_PARAMS
+);
+
+define_server_key_scalar_div_bench_fn!(
+    method_name: unchecked_scalar_div,
+    display_name: div,
+    &SERVER_KEY_BENCH_PARAMS_EXTENDED
+);
+define_server_key_scalar_div_bench_fn!(
+    method_name: unchecked_scalar_mod,
+    display_name: modulo,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_add,
+    display_name: add,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_sub,
+    display_name: sub,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_mul,
+    display_name: mul,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_left_shift,
+    display_name: left_shift,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_right_shift,
+    display_name: right_shift,
+    &SERVER_KEY_BENCH_PARAMS
+);
+
+define_server_key_scalar_div_bench_fn!(
+    method_name: scalar_div,
+    display_name: div,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_div_bench_fn!(
+    method_name: scalar_mod,
+    display_name: modulo,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_greater,
+    display_name: greater,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_greater_or_equal,
+    display_name: greater_or_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_less,
+    display_name: less,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_bench_fn!(
+    method_name: scalar_less_or_equal,
+    display_name: less_or_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_div_bench_fn!(
+    method_name: scalar_equal,
+    display_name: equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
+define_server_key_scalar_div_bench_fn!(
+    method_name: scalar_not_equal,
+    display_name: not_equal,
+    &SERVER_KEY_BENCH_PARAMS
+);
 
 criterion_group!(
-    arithmetic_operation,
-    unchecked_add,
-    unchecked_sub,
-    unchecked_mul_lsb,
-    unchecked_mul_msb,
+    smart_ops,
     smart_bitand,
     smart_bitor,
     smart_bitxor,
     smart_add,
     smart_sub,
     smart_mul_lsb,
+);
+
+criterion_group!(
+    unchecked_ops,
+    unchecked_neg,
+    unchecked_add,
+    unchecked_sub,
+    unchecked_mul_lsb,
+    unchecked_mul_msb,
+    unchecked_div,
+    unchecked_greater,
+    unchecked_less,
+    unchecked_equal,
     carry_extract,
-    // programmable_bootstrapping,
-    // multivalue_programmable_bootstrapping
-    //bench_two_block_pbs
-    //wopbs_v0_norm2_2,
-    bench_wopbs_param_message_8_norm2_5,
     programmable_bootstrapping
 );
 
 criterion_group!(
-    arithmetic_scalar_operation,
+    unchecked_scalar_ops,
     unchecked_scalar_add,
     unchecked_scalar_mul,
+    unchecked_scalar_sub,
+    unchecked_scalar_div,
+    unchecked_scalar_mod,
+    unchecked_scalar_left_shift,
+    unchecked_scalar_right_shift,
 );
 
-criterion_main!(arithmetic_operation,); // arithmetic_scalar_operation,);
+criterion_group!(
+    default_ops,
+    neg,
+    bitand,
+    bitor,
+    bitxor,
+    add,
+    sub,
+    div,
+    mul,
+    greater,
+    greater_or_equal,
+    less,
+    less_or_equal,
+    equal,
+    not_equal
+);
+
+criterion_group!(
+    default_scalar_ops,
+    scalar_add,
+    scalar_sub,
+    scalar_div,
+    scalar_mul,
+    scalar_mod,
+    scalar_left_shift,
+    scalar_right_shift,
+    scalar_greater,
+    scalar_greater_or_equal,
+    scalar_less,
+    scalar_less_or_equal,
+    scalar_equal,
+    scalar_not_equal
+);
+
+mod casting;
+criterion_group!(
+    casting,
+    casting::pack_cast_64,
+    casting::pack_cast,
+    casting::cast
+);
+
+fn main() {
+    fn default_bench() {
+        casting();
+        default_ops();
+        default_scalar_ops();
+    }
+
+    match env::var("__TFHE_RS_BENCH_OP_FLAVOR") {
+        Ok(val) => {
+            match val.to_lowercase().as_str() {
+                "default" => default_bench(),
+                "smart" => smart_ops(),
+                "unchecked" => {
+                    unchecked_ops();
+                    unchecked_scalar_ops();
+                }
+                _ => panic!("unknown benchmark operations flavor"),
+            };
+        }
+        Err(_) => default_bench(),
+    };
+
+    Criterion::default().configure_from_args().final_summary();
+}

tfhe/benches/shortint/casting.rs

@@ -0,0 +1,137 @@
+use crate::utilities::{write_to_json, OperatorType};
+
+use tfhe::shortint::prelude::*;
+
+use rayon::prelude::*;
+
+use criterion::Criterion;
+
+pub fn pack_cast_64(c: &mut Criterion) {
+    let bench_name = "pack_cast_64";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    let (client_key_1, server_key_1): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+    let (client_key_2, server_key_2): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_2_CARRY_2_KS_PBS);
+
+    let ks_param = PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS;
+    let ks_param_name = "PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS";
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key_1, &server_key_1),
+        (&client_key_2, &server_key_2),
+        ks_param,
+    );
+
+    let vec_ct = vec![client_key_1.encrypt(1); 64];
+
+    let bench_id = format!("{bench_name}_{ks_param_name}");
+    bench_group.bench_function(&bench_id, |b| {
+        b.iter(|| {
+            let _ = (0..32)
+                .into_par_iter()
+                .map(|i| {
+                    let byte_idx = 7 - i / 4;
+                    let pair_idx = i % 4;
+
+                    let b0 = &vec_ct[8 * byte_idx + 2 * pair_idx];
+                    let b1 = &vec_ct[8 * byte_idx + 2 * pair_idx + 1];
+
+                    ksk.cast(
+                        &server_key_1.unchecked_add(b0, &server_key_1.unchecked_scalar_mul(b1, 2)),
+                    )
+                })
+                .collect::<Vec<_>>();
+        });
+    });
+
+    write_to_json::<u64, _>(
+        &bench_id,
+        ks_param,
+        ks_param_name,
+        "pack_cast_64",
+        &OperatorType::Atomic,
+        0,
+        vec![],
+    );
+}
+
+pub fn pack_cast(c: &mut Criterion) {
+    let bench_name = "pack_cast";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    let (client_key_1, server_key_1): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+    let (client_key_2, server_key_2): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_2_CARRY_2_KS_PBS);
+
+    let ks_param = PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS;
+    let ks_param_name = "PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS";
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key_1, &server_key_1),
+        (&client_key_2, &server_key_2),
+        ks_param,
+    );
+
+    let ct_1 = client_key_1.encrypt(1);
+    let ct_2 = client_key_1.encrypt(1);
+
+    let bench_id = format!("{bench_name}_{ks_param_name}");
+    bench_group.bench_function(&bench_id, |b| {
+        b.iter(|| {
+            let _ = ksk.cast(
+                &server_key_1.unchecked_add(&ct_1, &server_key_1.unchecked_scalar_mul(&ct_2, 2)),
+            );
+        });
+    });
+
+    write_to_json::<u64, _>(
+        &bench_id,
+        ks_param,
+        ks_param_name,
+        "pack_cast",
+        &OperatorType::Atomic,
+        0,
+        vec![],
+    );
+}
+
+pub fn cast(c: &mut Criterion) {
+    let bench_name = "cast";
+    let mut bench_group = c.benchmark_group(bench_name);
+
+    let (client_key_1, server_key_1): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_1_CARRY_1_KS_PBS);
+    let (client_key_2, server_key_2): (ClientKey, ServerKey) =
+        gen_keys(PARAM_MESSAGE_2_CARRY_2_KS_PBS);
+
+    let ks_param = PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS;
+    let ks_param_name = "PARAM_KEYSWITCH_1_1_KS_PBS_TO_2_2_KS_PBS";
+
+    let ksk = KeySwitchingKey::new(
+        (&client_key_1, &server_key_1),
+        (&client_key_2, &server_key_2),
+        ks_param,
+    );
+
+    let ct = client_key_1.encrypt(1);
+
+    let bench_id = format!("{bench_name}_{ks_param_name}");
+    bench_group.bench_function(&bench_id, |b| {
+        b.iter(|| {
+            let _ = ksk.cast(&ct);
+        });
+    });
+
+    write_to_json::<u64, _>(
+        &bench_id,
+        ks_param,
+        ks_param_name,
+        "cast",
+        &OperatorType::Atomic,
+        0,
+        vec![],
+    );
+}

tfhe/benches/utilities.rs

@@ -0,0 +1,231 @@
+use serde::Serialize;
+use std::fs;
+use std::path::PathBuf;
+#[cfg(feature = "boolean")]
+use tfhe::boolean::parameters::BooleanParameters;
+use tfhe::core_crypto::prelude::*;
+#[cfg(feature = "shortint")]
+use tfhe::shortint::parameters::ShortintKeySwitchingParameters;
+#[cfg(feature = "shortint")]
+use tfhe::shortint::PBSParameters;
+
+#[derive(Clone, Copy, Default, Serialize)]
+pub struct CryptoParametersRecord<Scalar: UnsignedInteger> {
+    pub lwe_dimension: Option<LweDimension>,
+    pub glwe_dimension: Option<GlweDimension>,
+    pub polynomial_size: Option<PolynomialSize>,
+    pub lwe_modular_std_dev: Option<StandardDev>,
+    pub glwe_modular_std_dev: Option<StandardDev>,
+    pub pbs_base_log: Option<DecompositionBaseLog>,
+    pub pbs_level: Option<DecompositionLevelCount>,
+    pub ks_base_log: Option<DecompositionBaseLog>,
+    pub ks_level: Option<DecompositionLevelCount>,
+    pub pfks_level: Option<DecompositionLevelCount>,
+    pub pfks_base_log: Option<DecompositionBaseLog>,
+    pub pfks_modular_std_dev: Option<StandardDev>,
+    pub cbs_level: Option<DecompositionLevelCount>,
+    pub cbs_base_log: Option<DecompositionBaseLog>,
+    pub message_modulus: Option<usize>,
+    pub carry_modulus: Option<usize>,
+    pub ciphertext_modulus: Option<CiphertextModulus<Scalar>>,
+}
+
+#[cfg(feature = "boolean")]
+impl<Scalar: UnsignedInteger> From<BooleanParameters> for CryptoParametersRecord<Scalar> {
+    fn from(params: BooleanParameters) -> Self {
+        CryptoParametersRecord {
+            lwe_dimension: Some(params.lwe_dimension),
+            glwe_dimension: Some(params.glwe_dimension),
+            polynomial_size: Some(params.polynomial_size),
+            lwe_modular_std_dev: Some(params.lwe_modular_std_dev),
+            glwe_modular_std_dev: Some(params.glwe_modular_std_dev),
+            pbs_base_log: Some(params.pbs_base_log),
+            pbs_level: Some(params.pbs_level),
+            ks_base_log: Some(params.ks_base_log),
+            ks_level: Some(params.ks_level),
+            pfks_level: None,
+            pfks_base_log: None,
+            pfks_modular_std_dev: None,
+            cbs_level: None,
+            cbs_base_log: None,
+            message_modulus: None,
+            carry_modulus: None,
+            ciphertext_modulus: Some(CiphertextModulus::<Scalar>::new_native()),
+        }
+    }
+}
+
+#[cfg(feature = "shortint")]
+impl<Scalar> From<PBSParameters> for CryptoParametersRecord<Scalar>
+where
+    Scalar: UnsignedInteger + CastInto<u128>,
+{
+    fn from(params: PBSParameters) -> Self {
+        CryptoParametersRecord {
+            lwe_dimension: Some(params.lwe_dimension()),
+            glwe_dimension: Some(params.glwe_dimension()),
+            polynomial_size: Some(params.polynomial_size()),
+            lwe_modular_std_dev: Some(params.lwe_modular_std_dev()),
+            glwe_modular_std_dev: Some(params.glwe_modular_std_dev()),
+            pbs_base_log: Some(params.pbs_base_log()),
+            pbs_level: Some(params.pbs_level()),
+            ks_base_log: Some(params.ks_base_log()),
+            ks_level: Some(params.ks_level()),
+            pfks_level: None,
+            pfks_base_log: None,
+            pfks_modular_std_dev: None,
+            cbs_level: None,
+            cbs_base_log: None,
+            message_modulus: Some(params.message_modulus().0),
+            carry_modulus: Some(params.carry_modulus().0),
+            ciphertext_modulus: Some(
+                params
+                    .ciphertext_modulus()
+                    .try_to()
+                    .expect("failed to convert ciphertext modulus"),
+            ),
+        }
+    }
+}
+
+#[cfg(feature = "shortint")]
+impl<Scalar: UnsignedInteger> From<ShortintKeySwitchingParameters>
+    for CryptoParametersRecord<Scalar>
+{
+    fn from(params: ShortintKeySwitchingParameters) -> Self {
+        CryptoParametersRecord {
+            lwe_dimension: None,
+            glwe_dimension: None,
+            polynomial_size: None,
+            lwe_modular_std_dev: None,
+            glwe_modular_std_dev: None,
+            pbs_base_log: None,
+            pbs_level: None,
+            ks_base_log: Some(params.ks_base_log),
+            ks_level: Some(params.ks_level),
+            pfks_level: None,
+            pfks_base_log: None,
+            pfks_modular_std_dev: None,
+            cbs_level: None,
+            cbs_base_log: None,
+            message_modulus: None,
+            carry_modulus: None,
+            ciphertext_modulus: None,
+        }
+    }
+}
+
+#[derive(Serialize)]
+enum PolynomialMultiplication {
+    Fft,
+    // Ntt,
+}
+
+#[derive(Serialize)]
+enum IntegerRepresentation {
+    Radix,
+    // Crt,
+    // Hybrid,
+}
+
+#[derive(Serialize)]
+enum ExecutionType {
+    Sequential,
+    Parallel,
+}
+
+#[derive(Serialize)]
+enum KeySetType {
+    Single,
+    // Multi,
+}
+
+#[derive(Serialize)]
+enum OperandType {
+    CipherText,
+    PlainText,
+}
+
+#[derive(Clone, Serialize)]
+pub enum OperatorType {
+    Atomic,
+    // AtomicPattern,
+}
+
+#[derive(Serialize)]
+struct BenchmarkParametersRecord<Scalar: UnsignedInteger> {
+    display_name: String,
+    crypto_parameters_alias: String,
+    crypto_parameters: CryptoParametersRecord<Scalar>,
+    message_modulus: Option<usize>,
+    carry_modulus: Option<usize>,
+    ciphertext_modulus: usize,
+    bit_size: u32,
+    polynomial_multiplication: PolynomialMultiplication,
+    precision: u32,
+    error_probability: f64,
+    integer_representation: IntegerRepresentation,
+    decomposition_basis: Vec<u32>,
+    pbs_algorithm: Option<String>,
+    execution_type: ExecutionType,
+    key_set_type: KeySetType,
+    operand_type: OperandType,
+    operator_type: OperatorType,
+}
+
+/// Writes benchmarks parameters to disk in JSON format.
+pub fn write_to_json<
+    Scalar: UnsignedInteger + Serialize,
+    T: Into<CryptoParametersRecord<Scalar>>,
+>(
+    bench_id: &str,
+    params: T,
+    params_alias: impl Into<String>,
+    display_name: impl Into<String>,
+    operator_type: &OperatorType,
+    bit_size: u32,
+    decomposition_basis: Vec<u32>,
+) {
+    let params = params.into();
+
+    let execution_type = match bench_id.contains("parallelized") {
+        true => ExecutionType::Parallel,
+        false => ExecutionType::Sequential,
+    };
+    let operand_type = match bench_id.contains("scalar") {
+        true => OperandType::PlainText,
+        false => OperandType::CipherText,
+    };
+
+    let record = BenchmarkParametersRecord {
+        display_name: display_name.into(),
+        crypto_parameters_alias: params_alias.into(),
+        crypto_parameters: params.to_owned(),
+        message_modulus: params.message_modulus,
+        carry_modulus: params.carry_modulus,
+        ciphertext_modulus: 64,
+        bit_size,
+        polynomial_multiplication: PolynomialMultiplication::Fft,
+        precision: (params.message_modulus.unwrap_or(2) as u32).ilog2(),
+        error_probability: 2f64.powf(-41.0),
+        integer_representation: IntegerRepresentation::Radix,
+        decomposition_basis,
+        pbs_algorithm: None, // To be added in future version
+        execution_type,
+        key_set_type: KeySetType::Single,
+        operand_type,
+        operator_type: operator_type.to_owned(),
+    };
+
+    let mut params_directory = ["benchmarks_parameters", bench_id]
+        .iter()
+        .collect::<PathBuf>();
+    fs::create_dir_all(&params_directory).unwrap();
+    params_directory.push("parameters.json");
+
+    fs::write(params_directory, serde_json::to_string(&record).unwrap()).unwrap();
+}
+
+// Empty main to please clippy.
+#[allow(dead_code)]
+pub fn main() {}

tfhe/build.rs

@@ -1,10 +1,24 @@
-// tfhe/build.rs
-
 #[cfg(feature = "__c_api")]
 fn gen_c_api() {
     use std::env;
     use std::path::PathBuf;
 
+    if std::env::var("_CBINDGEN_IS_RUNNING").is_ok() {
+        return;
+    }
+
+    fn get_build_profile_name() -> String {
+        // The profile name is always the 3rd last part of the path (with 1 based indexing).
+        // e.g. /code/core/target/cli/build/my-build-info-9f91ba6f99d7a061/out
+        let out_dir = std::env::var("OUT_DIR")
+            .expect("OUT_DIR is not set, cannot determine build profile, aborting");
+        out_dir
+            .split(std::path::MAIN_SEPARATOR)
+            .nth_back(3)
+            .expect("Cannot determine build profile, aborting")
+            .to_string()
+    }
+
     /// Find the location of the `target/` directory. Note that this may be
     /// overridden by `cmake`, so we also need to check the `CARGO_TARGET_DIR`
     /// variable.
@@ -12,7 +26,8 @@ fn gen_c_api() {
         if let Ok(target) = env::var("CARGO_TARGET_DIR") {
             PathBuf::from(target)
         } else {
-            PathBuf::from(env::var("CARGO_MANIFEST_DIR").unwrap()).join("../target/release")
+            PathBuf::from(env::var("CARGO_MANIFEST_DIR").unwrap())
+                .join(format!("../target/{}", get_build_profile_name()))
         }
     }
 
@@ -24,7 +39,35 @@ fn gen_c_api() {
         .display()
         .to_string();
 
-    cbindgen::generate(crate_dir)
+    let parse_expand_features_vec = vec![
+        #[cfg(feature = "__c_api")]
+        "__c_api",
+        #[cfg(feature = "boolean-c-api")]
+        "boolean-c-api",
+        #[cfg(feature = "shortint-c-api")]
+        "shortint-c-api",
+        #[cfg(feature = "high-level-c-api")]
+        "high-level-c-api",
+        #[cfg(feature = "boolean")]
+        "boolean",
+        #[cfg(feature = "shortint")]
+        "shortint",
+        #[cfg(feature = "integer")]
+        "integer",
+    ];
+
+    let parse_expand_vec = if parse_expand_features_vec.is_empty() {
+        vec![]
+    } else {
+        vec![package_name.as_str()]
+    };
+
+    cbindgen::Builder::new()
+        .with_crate(crate_dir.clone())
+        .with_config(cbindgen::Config::from_root_or_default(crate_dir))
+        .with_parse_expand(&parse_expand_vec)
+        .with_parse_expand_features(&parse_expand_features_vec)
+        .generate()
         .unwrap()
         .write_to_file(output_file);
 }

tfhe/c_api_tests/CMakeLists.txt

@@ -2,7 +2,10 @@ project(tfhe-c-api-tests)
 
 cmake_minimum_required(VERSION 3.16)
 
-set(TFHE_C_API_RELEASE "${CMAKE_CURRENT_SOURCE_DIR}/../../target/release/")
+if(NOT CARGO_PROFILE)
+    set(CARGO_PROFILE release)
+endif()
+set(TFHE_C_API_RELEASE "${CMAKE_CURRENT_SOURCE_DIR}/../../target/${CARGO_PROFILE}")
 
 include_directories(${TFHE_C_API_RELEASE})
 add_library(Tfhe STATIC IMPORTED)

tfhe/c_api_tests/test_boolean_keygen.c

@@ -61,13 +61,13 @@ void test_default_keygen_w_serde(void) {
 
   assert(c_result == true);
 
-  destroy_boolean_client_key(cks);
-  destroy_boolean_server_key(sks);
-  destroy_boolean_ciphertext(ct);
-  destroy_boolean_ciphertext(deser_ct);
-  destroy_boolean_compressed_ciphertext(cct);
-  destroy_boolean_compressed_ciphertext(deser_cct);
-  destroy_boolean_ciphertext(decompressed_ct);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_server_key(sks);
+  boolean_destroy_ciphertext(ct);
+  boolean_destroy_ciphertext(deser_ct);
+  boolean_destroy_compressed_ciphertext(cct);
+  boolean_destroy_compressed_ciphertext(deser_cct);
+  boolean_destroy_ciphertext(decompressed_ct);
   destroy_buffer(&ct_ser_buffer);
 }
 
@@ -75,50 +75,52 @@ void test_predefined_keygen_w_serde(void) {
   BooleanClientKey *cks = NULL;
   BooleanServerKey *sks = NULL;
 
-  int gen_keys_ok = boolean_gen_keys_with_predefined_parameters_set(
-      BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &cks, &sks);
+  int gen_keys_ok =
+      boolean_gen_keys_with_parameters(BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &cks, &sks);
 
   assert(gen_keys_ok == 0);
 
-  destroy_boolean_client_key(cks);
-  destroy_boolean_server_key(sks);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_server_key(sks);
 
-  gen_keys_ok = boolean_gen_keys_with_predefined_parameters_set(
-      BOOLEAN_PARAMETERS_SET_TFHE_LIB_PARAMETERS, &cks, &sks);
+  gen_keys_ok =
+      boolean_gen_keys_with_parameters(BOOLEAN_PARAMETERS_SET_TFHE_LIB_PARAMETERS, &cks, &sks);
 
   assert(gen_keys_ok == 0);
 
-  destroy_boolean_client_key(cks);
-  destroy_boolean_server_key(sks);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_server_key(sks);
 }
 
 void test_custom_keygen(void) {
   BooleanClientKey *cks = NULL;
   BooleanServerKey *sks = NULL;
-  BooleanParameters *params = NULL;
-
-  int params_ok = boolean_create_parameters(10, 1, 1024, 10e-100, 10e-100, 3, 1, 4, 2, &params);
-  assert(params_ok == 0);
+  BooleanParameters params = {
+      .lwe_dimension = 10,
+      .glwe_dimension = 1,
+      .polynomial_size = 1024,
+      .lwe_modular_std_dev = 10e-100,
+      .glwe_modular_std_dev = 10e-100,
+      .pbs_base_log = 3,
+      .pbs_level = 1,
+      .ks_base_log = 4,
+      .ks_level = 2,
+  };
 
   int gen_keys_ok = boolean_gen_keys_with_parameters(params, &cks, &sks);
 
   assert(gen_keys_ok == 0);
 
-  destroy_boolean_parameters(params);
-  destroy_boolean_client_key(cks);
-  destroy_boolean_server_key(sks);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_server_key(sks);
 }
 
 void test_public_keygen(void) {
   BooleanClientKey *cks = NULL;
   BooleanPublicKey *pks = NULL;
-  BooleanParameters *params = NULL;
   BooleanCiphertext *ct = NULL;
 
-  int get_params_ok = boolean_get_parameters(BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &params);
-  assert(get_params_ok == 0);
-
-  int gen_keys_ok = boolean_gen_client_key(params, &cks);
+  int gen_keys_ok = boolean_gen_client_key(BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &cks);
   assert(gen_keys_ok == 0);
 
   int gen_pks = boolean_gen_public_key(cks, &pks);
@@ -135,10 +137,9 @@ void test_public_keygen(void) {
 
   assert(result == true);
 
-  destroy_boolean_parameters(params);
-  destroy_boolean_client_key(cks);
-  destroy_boolean_public_key(pks);
-  destroy_boolean_ciphertext(ct);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_public_key(pks);
+  boolean_destroy_ciphertext(ct);
 }
 
 int main(void) {

tfhe/c_api_tests/test_boolean_server_key.c

@@ -51,9 +51,9 @@ void test_binary_boolean_function(BooleanClientKey *cks, BooleanServerKey *sks,
 
           assert(decrypted_result == expected);
 
-          destroy_boolean_ciphertext(ct_left);
-          destroy_boolean_ciphertext(ct_right);
-          destroy_boolean_ciphertext(ct_result);
+          boolean_destroy_ciphertext(ct_left);
+          boolean_destroy_ciphertext(ct_right);
+          boolean_destroy_ciphertext(ct_result);
         }
       }
     }
@@ -103,8 +103,8 @@ void test_binary_boolean_function_assign(
 
           assert(decrypted_result == expected);
 
-          destroy_boolean_ciphertext(ct_left_and_result);
-          destroy_boolean_ciphertext(ct_right);
+          boolean_destroy_ciphertext(ct_left_and_result);
+          boolean_destroy_ciphertext(ct_right);
         }
       }
     }
@@ -139,8 +139,8 @@ void test_binary_boolean_function_scalar(BooleanClientKey *cks, BooleanServerKey
 
       assert(decrypted_result == expected);
 
-      destroy_boolean_ciphertext(ct_left);
-      destroy_boolean_ciphertext(ct_result);
+      boolean_destroy_ciphertext(ct_left);
+      boolean_destroy_ciphertext(ct_result);
     }
   }
 }
@@ -171,7 +171,7 @@ void test_binary_boolean_function_scalar_assign(BooleanClientKey *cks, BooleanSe
 
       assert(decrypted_result == expected);
 
-      destroy_boolean_ciphertext(ct_left_and_result);
+      boolean_destroy_ciphertext(ct_left_and_result);
     }
   }
 }
@@ -205,8 +205,8 @@ void test_not(BooleanClientKey *cks, BooleanServerKey *sks) {
 
       assert(decrypted_result == expected);
 
-      destroy_boolean_ciphertext(ct_in);
-      destroy_boolean_ciphertext(ct_result);
+      boolean_destroy_ciphertext(ct_in);
+      boolean_destroy_ciphertext(ct_result);
     }
   }
 }
@@ -239,7 +239,7 @@ void test_not_assign(BooleanClientKey *cks, BooleanServerKey *sks) {
 
       assert(decrypted_result == expected);
 
-      destroy_boolean_ciphertext(ct_in_and_result);
+      boolean_destroy_ciphertext(ct_in_and_result);
     }
   }
 }
@@ -300,10 +300,10 @@ void test_mux(BooleanClientKey *cks, BooleanServerKey *sks) {
 
               assert(decrypted_result == expected);
 
-              destroy_boolean_ciphertext(ct_cond);
-              destroy_boolean_ciphertext(ct_then);
-              destroy_boolean_ciphertext(ct_else);
-              destroy_boolean_ciphertext(ct_result);
+              boolean_destroy_ciphertext(ct_cond);
+              boolean_destroy_ciphertext(ct_then);
+              boolean_destroy_ciphertext(ct_else);
+              boolean_destroy_ciphertext(ct_result);
             }
           }
         }
@@ -334,12 +334,8 @@ void test_server_key(void) {
   BooleanCompressedServerKey *deser_csks = NULL;
   Buffer sks_ser_buffer = {.pointer = NULL, .length = 0};
   BooleanServerKey *deser_sks = NULL;
-  BooleanParameters *params = NULL;
 
-  int get_params_ok = boolean_get_parameters(BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &params);
-  assert(get_params_ok == 0);
-
-  int gen_cks_ok = boolean_gen_client_key(params, &cks);
+  int gen_cks_ok = boolean_gen_client_key(BOOLEAN_PARAMETERS_SET_DEFAULT_PARAMETERS, &cks);
   assert(gen_cks_ok == 0);
 
   int gen_csks_ok = boolean_gen_compressed_server_key(cks, &csks);
@@ -411,13 +407,12 @@ void test_server_key(void) {
   test_binary_boolean_function_scalar_assign(deser_cks, deser_sks, c_xnor,
                                              boolean_server_key_xnor_scalar_assign);
 
-  destroy_boolean_client_key(cks);
-  destroy_boolean_compressed_server_key(csks);
-  destroy_boolean_server_key(sks);
-  destroy_boolean_client_key(deser_cks);
-  destroy_boolean_compressed_server_key(deser_csks);
-  destroy_boolean_server_key(deser_sks);
-  destroy_boolean_parameters(params);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_compressed_server_key(csks);
+  boolean_destroy_server_key(sks);
+  boolean_destroy_client_key(deser_cks);
+  boolean_destroy_compressed_server_key(deser_csks);
+  boolean_destroy_server_key(deser_sks);
   destroy_buffer(&cks_ser_buffer);
   destroy_buffer(&csks_ser_buffer);
   destroy_buffer(&sks_ser_buffer);

tfhe/c_api_tests/test_high_level_128_bits.c

@@ -0,0 +1,123 @@
+#include <tfhe.h>
+
+#include <assert.h>
+#include <inttypes.h>
+#include <stdio.h>
+
+int uint128_client_key(const ClientKey *client_key) {
+  int ok;
+  FheUint128 *lhs = NULL;
+  FheUint128 *rhs = NULL;
+  FheUint128 *result = NULL;
+  U128 lhs_clear = {10, 20};
+  U128 rhs_clear = {1, 2};
+  U128 result_clear = {0};
+
+  ok = fhe_uint128_try_encrypt_with_client_key_u128(lhs_clear, client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_try_encrypt_with_client_key_u128(rhs_clear, client_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_sub(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint128_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 9);
+  assert(result_clear.w1 == 18);
+
+  fhe_uint128_destroy(lhs);
+  fhe_uint128_destroy(rhs);
+  fhe_uint128_destroy(result);
+  return ok;
+}
+
+int uint128_encrypt_trivial(const ClientKey *client_key) {
+  int ok;
+  FheUint128 *lhs = NULL;
+  FheUint128 *rhs = NULL;
+  FheUint128 *result = NULL;
+  U128 lhs_clear = {10, 20};
+  U128 rhs_clear = {1, 2};
+  U128 result_clear = {0};
+
+  ok = fhe_uint128_try_encrypt_trivial_u128(lhs_clear, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_try_encrypt_trivial_u128(rhs_clear, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_sub(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint128_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 9);
+  assert(result_clear.w1 == 18);
+
+  fhe_uint128_destroy(lhs);
+  fhe_uint128_destroy(rhs);
+  fhe_uint128_destroy(result);
+  return ok;
+}
+
+int uint128_public_key(const ClientKey *client_key, const PublicKey *public_key) {
+  int ok;
+  FheUint128 *lhs = NULL;
+  FheUint128 *rhs = NULL;
+  FheUint128 *result = NULL;
+  U128 lhs_clear = {10, 20};
+  U128 rhs_clear = {1, 2};
+  U128 result_clear = {0};
+
+  ok = fhe_uint128_try_encrypt_with_public_key_u128(lhs_clear, public_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_try_encrypt_with_public_key_u128(rhs_clear, public_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint128_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint128_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 11);
+  assert(result_clear.w1 == 22);
+
+  fhe_uint128_destroy(lhs);
+  fhe_uint128_destroy(rhs);
+  fhe_uint128_destroy(result);
+  return ok;
+}
+
+int main(void) {
+  int ok = 0;
+  ConfigBuilder *builder;
+  Config *config;
+
+  config_builder_all_disabled(&builder);
+  config_builder_enable_default_integers_small(&builder);
+  config_builder_build(builder, &config);
+
+  ClientKey *client_key = NULL;
+  ServerKey *server_key = NULL;
+  PublicKey *public_key = NULL;
+
+  generate_keys(config, &client_key, &server_key);
+  public_key_new(client_key, &public_key);
+
+  set_server_key(server_key);
+
+  uint128_client_key(client_key);
+  uint128_encrypt_trivial(client_key);
+  uint128_public_key(client_key, public_key);
+
+  client_key_destroy(client_key);
+  public_key_destroy(public_key);
+  server_key_destroy(server_key);
+  return ok;
+}

tfhe/c_api_tests/test_high_level_256_bits.c

@@ -0,0 +1,139 @@
+#include <tfhe.h>
+
+#include <assert.h>
+#include <inttypes.h>
+#include <stdio.h>
+
+int uint256_client_key(const ClientKey *client_key) {
+  int ok;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  FheUint64 *cast_result = NULL;
+  U256 lhs_clear = {1, 2, 3, 4};
+  U256 rhs_clear = {5, 6, 7, 8};
+  U256 result_clear = {0};
+
+  ok = fhe_uint256_try_encrypt_with_client_key_u256(lhs_clear, client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_try_encrypt_with_client_key_u256(rhs_clear, client_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 6);
+  assert(result_clear.w1 == 8);
+  assert(result_clear.w2 == 10);
+  assert(result_clear.w3 == 12);
+
+  // try some casting
+  ok = fhe_uint256_cast_into_fhe_uint64(result, &cast_result);
+  assert(ok == 0);
+  uint64_t u64_clear;
+  ok = fhe_uint64_decrypt(cast_result, client_key, &u64_clear);
+  assert(ok == 0);
+  assert(u64_clear == 6);
+
+  fhe_uint256_destroy(lhs);
+  fhe_uint256_destroy(rhs);
+  fhe_uint256_destroy(result);
+  fhe_uint64_destroy(cast_result);
+  return ok;
+}
+
+int uint256_encrypt_trivial(const ClientKey *client_key) {
+  int ok;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  U256 lhs_clear = {1, 2, 3, 4};
+  U256 rhs_clear = {5, 6, 7, 8};
+  U256 result_clear = {0};
+
+  ok = fhe_uint256_try_encrypt_trivial_u256(lhs_clear, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_try_encrypt_trivial_u256(rhs_clear, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 6);
+  assert(result_clear.w1 == 8);
+  assert(result_clear.w2 == 10);
+  assert(result_clear.w3 == 12);
+
+  fhe_uint256_destroy(lhs);
+  fhe_uint256_destroy(rhs);
+  fhe_uint256_destroy(result);
+  return ok;
+}
+
+int uint256_public_key(const ClientKey *client_key, const PublicKey *public_key) {
+  int ok;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  U256 lhs_clear = {5, 6, 7, 8};
+  U256 rhs_clear = {1, 2, 3, 4};
+  U256 result_clear = {0};
+
+  ok = fhe_uint256_try_encrypt_with_public_key_u256(lhs_clear, public_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_try_encrypt_with_public_key_u256(rhs_clear, public_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_sub(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 4);
+  assert(result_clear.w1 == 4);
+  assert(result_clear.w2 == 4);
+  assert(result_clear.w3 == 4);
+
+  fhe_uint256_destroy(lhs);
+  fhe_uint256_destroy(rhs);
+  fhe_uint256_destroy(result);
+  return ok;
+}
+
+int main(void) {
+  int ok = 0;
+  ConfigBuilder *builder;
+  Config *config;
+
+  config_builder_all_disabled(&builder);
+  config_builder_enable_default_integers_small(&builder);
+  config_builder_build(builder, &config);
+
+  ClientKey *client_key = NULL;
+  ServerKey *server_key = NULL;
+  PublicKey *public_key = NULL;
+
+  generate_keys(config, &client_key, &server_key);
+  public_key_new(client_key, &public_key);
+
+  set_server_key(server_key);
+
+  uint256_client_key(client_key);
+  uint256_encrypt_trivial(client_key);
+  uint256_public_key(client_key, public_key);
+
+  client_key_destroy(client_key);
+  public_key_destroy(public_key);
+  server_key_destroy(server_key);
+  return ok;
+}

tfhe/c_api_tests/test_high_level_boolean.c

@@ -0,0 +1,127 @@
+#include <tfhe.h>
+
+#include <assert.h>
+#include <inttypes.h>
+#include <stdio.h>
+
+int client_key_test(const ClientKey *client_key) {
+  int ok;
+  FheBool *lhs = NULL;
+  FheBool *rhs = NULL;
+  FheBool *result = NULL;
+
+  bool lhs_clear = 0;
+  bool rhs_clear = 1;
+
+  ok = fhe_bool_try_encrypt_with_client_key_bool(lhs_clear, client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_try_encrypt_with_client_key_bool(rhs_clear, client_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_bitand(lhs, rhs, &result);
+  assert(ok == 0);
+
+  bool clear;
+  ok = fhe_bool_decrypt(result, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == (lhs_clear & rhs_clear));
+
+  fhe_bool_destroy(lhs);
+  fhe_bool_destroy(rhs);
+  fhe_bool_destroy(result);
+
+  return ok;
+}
+
+int public_key_test(const ClientKey *client_key, const PublicKey *public_key) {
+  int ok;
+  FheBool *lhs = NULL;
+  FheBool *rhs = NULL;
+  FheBool *result = NULL;
+
+  bool lhs_clear = 0;
+  bool rhs_clear = 1;
+
+  ok = fhe_bool_try_encrypt_with_public_key_bool(lhs_clear, public_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_try_encrypt_with_public_key_bool(rhs_clear, public_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_bitand(lhs, rhs, &result);
+  assert(ok == 0);
+
+  bool clear;
+  ok = fhe_bool_decrypt(result, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == (lhs_clear & rhs_clear));
+
+  fhe_bool_destroy(lhs);
+  fhe_bool_destroy(rhs);
+  fhe_bool_destroy(result);
+
+  return ok;
+}
+
+int trivial_encrypt_test(const ClientKey *client_key) {
+  int ok;
+  FheBool *lhs = NULL;
+  FheBool *rhs = NULL;
+  FheBool *result = NULL;
+
+  bool lhs_clear = 0;
+  bool rhs_clear = 1;
+
+  ok = fhe_bool_try_encrypt_trivial_bool(lhs_clear, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_try_encrypt_trivial_bool(rhs_clear, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_bool_bitand(lhs, rhs, &result);
+  assert(ok == 0);
+
+  bool clear;
+  ok = fhe_bool_decrypt(result, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == (lhs_clear & rhs_clear));
+
+  fhe_bool_destroy(lhs);
+  fhe_bool_destroy(rhs);
+  fhe_bool_destroy(result);
+
+  return ok;
+}
+
+int main(void) {
+
+  ConfigBuilder *builder;
+  Config *config;
+
+  config_builder_all_disabled(&builder);
+  config_builder_enable_default_bool(&builder);
+  config_builder_build(builder, &config);
+
+  ClientKey *client_key = NULL;
+  ServerKey *server_key = NULL;
+  PublicKey *public_key = NULL;
+
+  generate_keys(config, &client_key, &server_key);
+  public_key_new(client_key, &public_key);
+
+  set_server_key(server_key);
+
+  client_key_test(client_key);
+  public_key_test(client_key, public_key);
+  trivial_encrypt_test(client_key);
+
+  client_key_destroy(client_key);
+  public_key_destroy(public_key);
+  server_key_destroy(server_key);
+
+  return EXIT_SUCCESS;
+}

tfhe/c_api_tests/test_high_level_custom_integers.c

@@ -0,0 +1,217 @@
+#include <tfhe.h>
+
+#include <assert.h>
+#include <inttypes.h>
+#include <stdio.h>
+
+int uint256_client_key(const ClientKey *client_key) {
+  int ok;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  FheUint64 *cast_result = NULL;
+  U256 lhs_clear = {1, 2, 3, 4};
+  U256 rhs_clear = {5, 6, 7, 8};
+  U256 result_clear = {0};
+
+  ok = fhe_uint256_try_encrypt_with_client_key_u256(lhs_clear, client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_try_encrypt_with_client_key_u256(rhs_clear, client_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 6);
+  assert(result_clear.w1 == 8);
+  assert(result_clear.w2 == 10);
+  assert(result_clear.w3 == 12);
+
+  // try some casting
+  ok = fhe_uint256_cast_into_fhe_uint64(result, &cast_result);
+  assert(ok == 0);
+  uint64_t u64_clear;
+  ok = fhe_uint64_decrypt(cast_result, client_key, &u64_clear);
+  assert(ok == 0);
+  assert(u64_clear == 6);
+
+  fhe_uint256_destroy(lhs);
+  fhe_uint256_destroy(rhs);
+  fhe_uint256_destroy(result);
+  fhe_uint64_destroy(cast_result);
+  return ok;
+}
+
+int uint256_encrypt_trivial(const ClientKey *client_key) {
+  int ok;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  U256 lhs_clear = {1, 2, 3, 4};
+  U256 rhs_clear = {5, 6, 7, 8};
+  U256 result_clear = {0};
+
+  ok = fhe_uint256_try_encrypt_trivial_u256(lhs_clear, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_try_encrypt_trivial_u256(rhs_clear, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint256_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+  assert(ok == 0);
+
+  assert(result_clear.w0 == 6);
+  assert(result_clear.w1 == 8);
+  assert(result_clear.w2 == 10);
+  assert(result_clear.w3 == 12);
+
+  fhe_uint256_destroy(lhs);
+  fhe_uint256_destroy(rhs);
+  fhe_uint256_destroy(result);
+  return ok;
+}
+
+int uint256_public_key(const ClientKey *client_key,
+                       const CompressedCompactPublicKey *compressed_public_key) {
+  int ok;
+  CompactPublicKey *public_key = NULL;
+  FheUint256 *lhs = NULL;
+  FheUint256 *rhs = NULL;
+  FheUint256 *result = NULL;
+  CompactFheUint256List *list = NULL;
+
+  U256 result_clear = {0};
+  U256 clears[2] = {{5, 6, 7, 8}, {1, 2, 3, 4}};
+
+  ok = compressed_compact_public_key_decompress(compressed_public_key, &public_key);
+  assert(ok == 0);
+
+  // Compact list example
+  {
+    ok = compact_fhe_uint256_list_try_encrypt_with_compact_public_key_u256(&clears[0], 2,
+                                                                           public_key, &list);
+    assert(ok == 0);
+
+    size_t len = 0;
+    ok = compact_fhe_uint256_list_len(list, &len);
+    assert(ok == 0);
+    assert(len == 2);
+
+    FheUint256 *expand_output[2] = {NULL};
+    ok = compact_fhe_uint256_list_expand(list, &expand_output[0], 2);
+    assert(ok == 0);
+
+    // transfer ownership
+    lhs = expand_output[0];
+    rhs = expand_output[1];
+    // We can destroy the compact list
+    // The expanded ciphertext are independant from it
+    compact_fhe_uint256_list_destroy(list);
+
+    ok = fhe_uint256_sub(lhs, rhs, &result);
+    assert(ok == 0);
+
+    ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+    assert(ok == 0);
+
+    assert(result_clear.w0 == 4);
+    assert(result_clear.w1 == 4);
+    assert(result_clear.w2 == 4);
+    assert(result_clear.w3 == 4);
+
+    fhe_uint256_destroy(lhs);
+    fhe_uint256_destroy(rhs);
+    fhe_uint256_destroy(result);
+  }
+
+  {
+    ok = fhe_uint256_try_encrypt_with_compact_public_key_u256(clears[0], public_key, &lhs);
+    assert(ok == 0);
+
+    ok = fhe_uint256_try_encrypt_with_compact_public_key_u256(clears[1], public_key, &rhs);
+    assert(ok == 0);
+
+    ok = fhe_uint256_sub(lhs, rhs, &result);
+    assert(ok == 0);
+
+    ok = fhe_uint256_decrypt(result, client_key, &result_clear);
+    assert(ok == 0);
+
+    assert(result_clear.w0 == 4);
+    assert(result_clear.w1 == 4);
+    assert(result_clear.w2 == 4);
+    assert(result_clear.w3 == 4);
+
+    fhe_uint256_destroy(lhs);
+    fhe_uint256_destroy(rhs);
+    fhe_uint256_destroy(result);
+  }
+
+  compact_public_key_destroy(public_key);
+  return ok;
+}
+
+int main(void) {
+  int ok = 0;
+  {
+    ConfigBuilder *builder;
+    Config *config;
+
+    config_builder_all_disabled(&builder);
+    config_builder_enable_custom_integers(&builder,
+                                          SHORTINT_PARAM_MESSAGE_2_CARRY_2_COMPACT_PK_KS_PBS);
+    config_builder_build(builder, &config);
+
+    ClientKey *client_key = NULL;
+    ServerKey *server_key = NULL;
+    CompressedCompactPublicKey *compressed_public_key = NULL;
+
+    generate_keys(config, &client_key, &server_key);
+    compressed_compact_public_key_new(client_key, &compressed_public_key);
+
+    set_server_key(server_key);
+
+    uint256_client_key(client_key);
+    uint256_encrypt_trivial(client_key);
+    uint256_public_key(client_key, compressed_public_key);
+
+    client_key_destroy(client_key);
+    compressed_compact_public_key_destroy(compressed_public_key);
+    server_key_destroy(server_key);
+  }
+
+  {
+    ConfigBuilder *builder;
+    Config *config;
+
+    config_builder_all_disabled(&builder);
+    config_builder_enable_custom_integers(&builder,
+                                          SHORTINT_PARAM_MESSAGE_2_CARRY_2_COMPACT_PK_PBS_KS);
+    config_builder_build(builder, &config);
+
+    ClientKey *client_key = NULL;
+    ServerKey *server_key = NULL;
+    CompressedCompactPublicKey *compressed_public_key = NULL;
+
+    generate_keys(config, &client_key, &server_key);
+    compressed_compact_public_key_new(client_key, &compressed_public_key);
+
+    set_server_key(server_key);
+
+    uint256_client_key(client_key);
+    uint256_encrypt_trivial(client_key);
+    uint256_public_key(client_key, compressed_public_key);
+
+    client_key_destroy(client_key);
+    compressed_compact_public_key_destroy(compressed_public_key);
+    server_key_destroy(server_key);
+  }
+  return ok;
+}

tfhe/c_api_tests/test_high_level_integers.c

@@ -0,0 +1,212 @@
+#include <tfhe.h>
+
+#include <assert.h>
+#include <inttypes.h>
+#include <stdio.h>
+
+int uint8_client_key(const ClientKey *client_key) {
+  int ok;
+  FheUint8 *lhs = NULL;
+  FheUint8 *rhs = NULL;
+  FheUint8 *result = NULL;
+
+  uint8_t lhs_clear = 123;
+  uint8_t rhs_clear = 14;
+
+  ok = fhe_uint8_try_encrypt_with_client_key_u8(lhs_clear, client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint8_try_encrypt_with_client_key_u8(rhs_clear, client_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint8_add(lhs, rhs, &result);
+  assert(ok == 0);
+
+  uint8_t clear;
+  ok = fhe_uint8_decrypt(result, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == (lhs_clear + rhs_clear));
+
+  fhe_uint8_destroy(lhs);
+  fhe_uint8_destroy(rhs);
+  fhe_uint8_destroy(result);
+  return ok;
+}
+
+int uint8_public_key(const ClientKey *client_key, const PublicKey *public_key) {
+  int ok;
+  FheUint8 *lhs = NULL;
+  FheUint8 *rhs = NULL;
+  FheUint8 *result = NULL;
+
+  uint8_t lhs_clear = 123;
+  uint8_t rhs_clear = 14;
+
+  ok = fhe_uint8_try_encrypt_with_public_key_u8(lhs_clear, public_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint8_try_encrypt_with_public_key_u8(rhs_clear, public_key, &rhs);
+  assert(ok == 0);
+
+  ok = fhe_uint8_sub(lhs, rhs, &result);
+  assert(ok == 0);
+
+  uint8_t clear;
+  ok = fhe_uint8_decrypt(result, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == (lhs_clear - rhs_clear));
+
+  fhe_uint8_destroy(lhs);
+  fhe_uint8_destroy(rhs);
+  fhe_uint8_destroy(result);
+  return ok;
+}
+
+int uint8_serialization(const ClientKey *client_key) {
+  int ok;
+  FheUint8 *lhs = NULL;
+  FheUint8 *deserialized_lhs = NULL;
+  FheUint8 *result = NULL;
+  Buffer value_buffer = {.pointer = NULL, .length = 0};
+  Buffer cks_buffer = {.pointer = NULL, .length = 0};
+  BufferView deser_view = {.pointer = NULL, .length = 0};
+  ClientKey *deserialized_client_key = NULL;
+
+  uint8_t lhs_clear = 123;
+
+  ok = client_key_serialize(client_key, &cks_buffer);
+  assert(ok == 0);
+
+  deser_view.pointer = cks_buffer.pointer;
+  deser_view.length = cks_buffer.length;
+  ok = client_key_deserialize(deser_view, &deserialized_client_key);
+  assert(ok == 0);
+
+  ok = fhe_uint8_try_encrypt_with_client_key_u8(lhs_clear, deserialized_client_key, &lhs);
+  assert(ok == 0);
+
+  ok = fhe_uint8_serialize(lhs, &value_buffer);
+  assert(ok == 0);
+
+  deser_view.pointer = value_buffer.pointer;
+  deser_view.length = value_buffer.length;
+  ok = fhe_uint8_deserialize(deser_view, &deserialized_lhs);
+  assert(ok == 0);
+
+  uint8_t clear;
+  ok = fhe_uint8_decrypt(deserialized_lhs, deserialized_client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == lhs_clear);
+
+  if (value_buffer.pointer != NULL) {
+    destroy_buffer(&value_buffer);
+  }
+  fhe_uint8_destroy(lhs);
+  fhe_uint8_destroy(deserialized_lhs);
+  fhe_uint8_destroy(result);
+  return ok;
+}
+
+int uint8_compressed(const ClientKey *client_key) {
+  int ok;
+  FheUint8 *lhs = NULL;
+  FheUint8 *result = NULL;
+  CompressedFheUint8 *clhs = NULL;
+
+  uint8_t lhs_clear = 123;
+
+  ok = compressed_fhe_uint8_try_encrypt_with_client_key_u8(lhs_clear, client_key, &clhs);
+  assert(ok == 0);
+
+  ok = compressed_fhe_uint8_decompress(clhs, &lhs);
+  assert(ok == 0);
+
+  uint8_t clear;
+  ok = fhe_uint8_decrypt(lhs, client_key, &clear);
+  assert(ok == 0);
+
+  assert(clear == lhs_clear);
+
+  fhe_uint8_destroy(lhs);
+  compressed_fhe_uint8_destroy(clhs);
+  fhe_uint8_destroy(result);
+  return ok;
+}
+
+int main(void) {
+  int ok = 0;
+  {
+    ConfigBuilder *builder;
+    Config *config;
+
+    ok = config_builder_all_disabled(&builder);
+    assert(ok == 0);
+    ok = config_builder_enable_default_integers(&builder);
+    assert(ok == 0);
+    ok = config_builder_build(builder, &config);
+    assert(ok == 0);
+
+    ClientKey *client_key = NULL;
+    ServerKey *server_key = NULL;
+    PublicKey *public_key = NULL;
+
+    ok = generate_keys(config, &client_key, &server_key);
+    assert(ok == 0);
+    ok = public_key_new(client_key, &public_key);
+    assert(ok == 0);
+    ok = uint8_serialization(client_key);
+    assert(ok == 0);
+    ok = uint8_compressed(client_key);
+    assert(ok == 0);
+
+    ok = set_server_key(server_key);
+    assert(ok == 0);
+
+    ok = uint8_client_key(client_key);
+    assert(ok == 0);
+    ok = uint8_public_key(client_key, public_key);
+    assert(ok == 0);
+
+    client_key_destroy(client_key);
+    public_key_destroy(public_key);
+    server_key_destroy(server_key);
+  }
+
+  {
+    ConfigBuilder *builder;
+    Config *config;
+
+    ok = config_builder_all_disabled(&builder);
+    assert(ok == 0);
+    ok = config_builder_enable_default_integers_small(&builder);
+    assert(ok == 0);
+    ok = config_builder_build(builder, &config);
+    assert(ok == 0);
+
+    ClientKey *client_key = NULL;
+    ServerKey *server_key = NULL;
+    PublicKey *public_key = NULL;
+
+    ok = generate_keys(config, &client_key, &server_key);
+    assert(ok == 0);
+    ok = public_key_new(client_key, &public_key);
+    assert(ok == 0);
+
+    ok = set_server_key(server_key);
+    assert(ok == 0);
+
+    ok = uint8_client_key(client_key);
+    assert(ok == 0);
+    ok = uint8_public_key(client_key, public_key);
+    assert(ok == 0);
+
+    client_key_destroy(client_key);
+    public_key_destroy(public_key);
+    server_key_destroy(server_key);
+  }
+
+  return ok;
+}

tfhe/c_api_tests/test_micro_bench_and.c

@@ -13,8 +13,8 @@ void micro_bench_and() {
   // int gen_keys_ok = boolean_gen_keys_with_default_parameters(&cks, &sks);
   // assert(gen_keys_ok == 0);
 
-  int gen_keys_ok = boolean_gen_keys_with_predefined_parameters_set(
-      BOOLEAN_PARAMETERS_SET_TFHE_LIB_PARAMETERS, &cks, &sks);
+  int gen_keys_ok =
+      boolean_gen_keys_with_parameters(BOOLEAN_PARAMETERS_SET_TFHE_LIB_PARAMETERS, &cks, &sks);
   assert(gen_keys_ok == 0);
 
   int num_loops = 10000;
@@ -32,7 +32,7 @@ void micro_bench_and() {
   for (int idx_loops = 0; idx_loops < num_loops; ++idx_loops) {
     BooleanCiphertext *ct_result = NULL;
     boolean_server_key_and(sks, ct_left, ct_right, &ct_result);
-    destroy_boolean_ciphertext(ct_result);
+    boolean_destroy_ciphertext(ct_result);
   }
 
   clock_t stop = clock();
@@ -41,8 +41,10 @@ void micro_bench_and() {
 
   printf("%g ms, mean %g ms\n", elapsed_ms, mean_ms);
 
-  destroy_boolean_client_key(cks);
-  destroy_boolean_server_key(sks);
+  boolean_destroy_client_key(cks);
+  boolean_destroy_server_key(sks);
+  boolean_destroy_ciphertext(ct_left);
+  boolean_destroy_ciphertext(ct_right);
 }
 
 int main(void) {

tfhe/c_api_tests/test_shortint_keygen.c

@@ -8,16 +8,13 @@
 void test_predefined_keygen_w_serde(void) {
   ShortintClientKey *cks = NULL;
   ShortintServerKey *sks = NULL;
-  ShortintParameters *params = NULL;
   ShortintCiphertext *ct = NULL;
   Buffer ct_ser_buffer = {.pointer = NULL, .length = 0};
   ShortintCiphertext *deser_ct = NULL;
   ShortintCompressedCiphertext *cct = NULL;
   ShortintCompressedCiphertext *deser_cct = NULL;
   ShortintCiphertext *decompressed_ct = NULL;
-
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
+  ShortintPBSParameters params = SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS;
 
   int gen_keys_ok = shortint_gen_keys_with_parameters(params, &cks, &sks);
   assert(gen_keys_ok == 0);
@@ -65,25 +62,21 @@ void test_predefined_keygen_w_serde(void) {
 
   assert(c_result == 3);
 
-  destroy_shortint_client_key(cks);
-  destroy_shortint_server_key(sks);
-  destroy_shortint_parameters(params);
-  destroy_shortint_ciphertext(ct);
-  destroy_shortint_ciphertext(deser_ct);
-  destroy_shortint_compressed_ciphertext(cct);
-  destroy_shortint_compressed_ciphertext(deser_cct);
-  destroy_shortint_ciphertext(decompressed_ct);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_server_key(sks);
+  shortint_destroy_ciphertext(ct);
+  shortint_destroy_ciphertext(deser_ct);
+  shortint_destroy_compressed_ciphertext(cct);
+  shortint_destroy_compressed_ciphertext(deser_cct);
+  shortint_destroy_ciphertext(decompressed_ct);
   destroy_buffer(&ct_ser_buffer);
 }
 
 void test_server_key_trivial_encrypt(void) {
   ShortintClientKey *cks = NULL;
   ShortintServerKey *sks = NULL;
-  ShortintParameters *params = NULL;
   ShortintCiphertext *ct = NULL;
-
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
+  ShortintPBSParameters params = SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS;
 
   int gen_keys_ok = shortint_gen_keys_with_parameters(params, &cks, &sks);
   assert(gen_keys_ok == 0);
@@ -97,41 +90,45 @@ void test_server_key_trivial_encrypt(void) {
 
   assert(result == 3);
 
-  destroy_shortint_client_key(cks);
-  destroy_shortint_server_key(sks);
-  destroy_shortint_parameters(params);
-  destroy_shortint_ciphertext(ct);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_server_key(sks);
+  shortint_destroy_ciphertext(ct);
 }
 
 void test_custom_keygen(void) {
   ShortintClientKey *cks = NULL;
   ShortintServerKey *sks = NULL;
-  ShortintParameters *params = NULL;
-
-  int params_ok = shortint_create_parameters(10, 1, 1024, 10e-100, 10e-100, 2, 3, 2, 3, 2, 3,
-                                             10e-100, 2, 3, 2, 2, &params);
-  assert(params_ok == 0);
+  ShortintPBSParameters params = {
+      .lwe_dimension = 10,
+      .glwe_dimension = 1,
+      .polynomial_size = 1024,
+      .lwe_modular_std_dev = 10e-100,
+      .glwe_modular_std_dev = 10e-100,
+      .pbs_base_log = 2,
+      .pbs_level = 3,
+      .ks_base_log = 2,
+      .ks_level = 3,
+      .message_modulus = 2,
+      .carry_modulus = 2,
+      .modulus_power_of_2_exponent = 64,
+      .encryption_key_choice = ShortintEncryptionKeyChoiceBig,
+  };
 
   int gen_keys_ok = shortint_gen_keys_with_parameters(params, &cks, &sks);
 
   assert(gen_keys_ok == 0);
 
-  destroy_shortint_parameters(params);
-  destroy_shortint_client_key(cks);
-  destroy_shortint_server_key(sks);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_server_key(sks);
 }
 
-void test_public_keygen(void) {
+void test_public_keygen(ShortintPBSParameters params) {
   ShortintClientKey *cks = NULL;
   ShortintPublicKey *pks = NULL;
   ShortintPublicKey *pks_deser = NULL;
-  ShortintParameters *params = NULL;
   ShortintCiphertext *ct = NULL;
   Buffer pks_ser_buff = {.pointer = NULL, .length = 0};
 
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
-
   int gen_keys_ok = shortint_gen_client_key(params, &cks);
   assert(gen_keys_ok == 0);
 
@@ -156,24 +153,19 @@ void test_public_keygen(void) {
 
   assert(result == 2);
 
-  destroy_shortint_parameters(params);
-  destroy_shortint_client_key(cks);
-  destroy_shortint_public_key(pks);
-  destroy_shortint_public_key(pks_deser);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_public_key(pks);
+  shortint_destroy_public_key(pks_deser);
   destroy_buffer(&pks_ser_buff);
-  destroy_shortint_ciphertext(ct);
+  shortint_destroy_ciphertext(ct);
 }
 
-void test_compressed_public_keygen(void) {
+void test_compressed_public_keygen(ShortintPBSParameters params) {
   ShortintClientKey *cks = NULL;
   ShortintCompressedPublicKey *cpks = NULL;
   ShortintPublicKey *pks = NULL;
-  ShortintParameters *params = NULL;
   ShortintCiphertext *ct = NULL;
 
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
-
   int gen_keys_ok = shortint_gen_client_key(params, &cks);
   assert(gen_keys_ok == 0);
 
@@ -203,18 +195,19 @@ void test_compressed_public_keygen(void) {
 
   assert(result == 2);
 
-  destroy_shortint_parameters(params);
-  destroy_shortint_client_key(cks);
-  destroy_shortint_compressed_public_key(cpks);
-  destroy_shortint_public_key(pks);
-  destroy_shortint_ciphertext(ct);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_compressed_public_key(cpks);
+  shortint_destroy_public_key(pks);
+  shortint_destroy_ciphertext(ct);
 }
 
 int main(void) {
   test_predefined_keygen_w_serde();
   test_custom_keygen();
-  test_public_keygen();
-  test_compressed_public_keygen();
+  test_public_keygen(SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS);
+  test_public_keygen(SHORTINT_PARAM_MESSAGE_2_CARRY_2_PBS_KS);
+  test_compressed_public_keygen(SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS);
+  test_compressed_public_keygen(SHORTINT_PARAM_MESSAGE_2_CARRY_2_PBS_KS);
   test_server_key_trivial_encrypt();
   return EXIT_SUCCESS;
 }

tfhe/c_api_tests/test_shortint_pbs.c

@@ -5,31 +5,31 @@
 #include <stdlib.h>
 #include <tgmath.h>
 
-uint64_t double_accumulator_2_bits_message(uint64_t in) { return (in * 2) % 4; }
+uint64_t double_lookup_table_2_bits_message(uint64_t in) { return (in * 2) % 4; }
 
-uint64_t get_max_value_of_accumulator_generator(uint64_t (*accumulator_func)(uint64_t),
-                                                size_t message_bits) {
+uint64_t get_max_value_of_lookup_table_generator(uint64_t (*lookup_table_func)(uint64_t),
+                                                 size_t message_bits) {
   uint64_t max_value = 0;
   for (size_t idx = 0; idx < (1 << message_bits); ++idx) {
-    uint64_t acc_value = accumulator_func((uint64_t)idx);
+    uint64_t acc_value = lookup_table_func((uint64_t)idx);
     max_value = acc_value > max_value ? acc_value : max_value;
   }
 
   return max_value;
 }
 
-uint64_t product_accumulator_2_bits_encrypted_mul(uint64_t left, uint64_t right) {
+uint64_t product_lookup_table_2_bits_encrypted_mul(uint64_t left, uint64_t right) {
   return (left * right) % 4;
 }
 
-uint64_t get_max_value_of_bivariate_accumulator_generator(uint64_t (*accumulator_func)(uint64_t,
-                                                                                       uint64_t),
-                                                          size_t message_bits_left,
-                                                          size_t message_bits_right) {
+uint64_t get_max_value_of_bivariate_lookup_table_generator(uint64_t (*lookup_table_func)(uint64_t,
+                                                                                         uint64_t),
+                                                           size_t message_bits_left,
+                                                           size_t message_bits_right) {
   uint64_t max_value = 0;
   for (size_t idx_left = 0; idx_left < (1 << message_bits_left); ++idx_left) {
     for (size_t idx_right = 0; idx_right < (1 << message_bits_right); ++idx_right) {
-      uint64_t acc_value = accumulator_func((uint64_t)idx_left, (uint64_t)idx_right);
+      uint64_t acc_value = lookup_table_func((uint64_t)idx_left, (uint64_t)idx_right);
       max_value = acc_value > max_value ? acc_value : max_value;
     }
   }
@@ -38,19 +38,16 @@ uint64_t get_max_value_of_bivariate_accumulator_generator(uint64_t (*accumulator
 }
 
 void test_shortint_pbs_2_bits_message(void) {
-  ShortintPBSAccumulator *accumulator = NULL;
+  ShortintPBSLookupTable *lookup_table = NULL;
   ShortintClientKey *cks = NULL;
   ShortintServerKey *sks = NULL;
-  ShortintParameters *params = NULL;
-
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
+  ShortintPBSParameters params = SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS;
 
   int gen_keys_ok = shortint_gen_keys_with_parameters(params, &cks, &sks);
   assert(gen_keys_ok == 0);
 
-  int gen_acc_ok = shortint_server_key_generate_pbs_accumulator(
-      sks, double_accumulator_2_bits_message, &accumulator);
+  int gen_acc_ok = shortint_server_key_generate_pbs_lookup_table(
+      sks, double_lookup_table_2_bits_message, &lookup_table);
   assert(gen_acc_ok == 0);
 
   for (int in_idx = 0; in_idx < 4; ++in_idx) {
@@ -68,11 +65,11 @@ void test_shortint_pbs_2_bits_message(void) {
 
     assert(degree == 3);
 
-    int pbs_ok = shortint_server_key_programmable_bootstrap(sks, accumulator, ct, &ct_out);
+    int pbs_ok = shortint_server_key_programmable_bootstrap(sks, lookup_table, ct, &ct_out);
     assert(pbs_ok == 0);
 
     size_t degree_to_set =
-        (size_t)get_max_value_of_accumulator_generator(double_accumulator_2_bits_message, 2);
+        (size_t)get_max_value_of_lookup_table_generator(double_lookup_table_2_bits_message, 2);
 
     int set_degree_ok = shortint_ciphertext_set_degree(ct_out, degree_to_set);
     assert(set_degree_ok == 0);
@@ -87,13 +84,14 @@ void test_shortint_pbs_2_bits_message(void) {
     int decrypt_non_assign_ok = shortint_client_key_decrypt(cks, ct_out, &result_non_assign);
     assert(decrypt_non_assign_ok == 0);
 
-    assert(result_non_assign == double_accumulator_2_bits_message(in_val));
+    assert(result_non_assign == double_lookup_table_2_bits_message(in_val));
 
-    int pbs_assign_ok = shortint_server_key_programmable_bootstrap_assign(sks, accumulator, ct_out);
+    int pbs_assign_ok =
+        shortint_server_key_programmable_bootstrap_assign(sks, lookup_table, ct_out);
     assert(pbs_assign_ok == 0);
 
     degree_to_set =
-        (size_t)get_max_value_of_accumulator_generator(double_accumulator_2_bits_message, 2);
+        (size_t)get_max_value_of_lookup_table_generator(double_lookup_table_2_bits_message, 2);
 
     set_degree_ok = shortint_ciphertext_set_degree(ct_out, degree_to_set);
     assert(set_degree_ok == 0);
@@ -102,32 +100,28 @@ void test_shortint_pbs_2_bits_message(void) {
     int decrypt_assign_ok = shortint_client_key_decrypt(cks, ct_out, &result_assign);
     assert(decrypt_assign_ok == 0);
 
-    assert(result_assign == double_accumulator_2_bits_message(result_non_assign));
+    assert(result_assign == double_lookup_table_2_bits_message(result_non_assign));
 
-    destroy_shortint_ciphertext(ct);
-    destroy_shortint_ciphertext(ct_out);
+    shortint_destroy_ciphertext(ct);
+    shortint_destroy_ciphertext(ct_out);
   }
 
-  destroy_shortint_pbs_accumulator(accumulator);
-  destroy_shortint_client_key(cks);
-  destroy_shortint_server_key(sks);
-  destroy_shortint_parameters(params);
+  shortint_destroy_pbs_lookup_table(lookup_table);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_server_key(sks);
 }
 
 void test_shortint_bivariate_pbs_2_bits_message(void) {
-  ShortintBivariatePBSAccumulator *accumulator = NULL;
+  ShortintBivariatePBSLookupTable *lookup_table = NULL;
   ShortintClientKey *cks = NULL;
   ShortintServerKey *sks = NULL;
-  ShortintParameters *params = NULL;
-
-  int get_params_ok = shortint_get_parameters(2, 2, &params);
-  assert(get_params_ok == 0);
+  ShortintPBSParameters params = SHORTINT_PARAM_MESSAGE_2_CARRY_2_KS_PBS;
 
   int gen_keys_ok = shortint_gen_keys_with_parameters(params, &cks, &sks);
   assert(gen_keys_ok == 0);
 
-  int gen_acc_ok = shortint_server_key_generate_bivariate_pbs_accumulator(
-      sks, product_accumulator_2_bits_encrypted_mul, &accumulator);
+  int gen_acc_ok = shortint_server_key_generate_bivariate_pbs_lookup_table(
+      sks, product_lookup_table_2_bits_encrypted_mul, &lookup_table);
   assert(gen_acc_ok == 0);
 
   for (int left_idx = 0; left_idx < 4; ++left_idx) {
@@ -145,12 +139,12 @@ void test_shortint_bivariate_pbs_2_bits_message(void) {
       int encrypt_right_ok = shortint_client_key_encrypt(cks, right_val, &ct_right);
       assert(encrypt_right_ok == 0);
 
-      int pbs_ok = shortint_server_key_bivariate_programmable_bootstrap(sks, accumulator, ct_left,
+      int pbs_ok = shortint_server_key_bivariate_programmable_bootstrap(sks, lookup_table, ct_left,
                                                                         ct_right, &ct_out);
       assert(pbs_ok == 0);
 
-      size_t degree_to_set = (size_t)get_max_value_of_bivariate_accumulator_generator(
-          product_accumulator_2_bits_encrypted_mul, 2, 2);
+      size_t degree_to_set = (size_t)get_max_value_of_bivariate_lookup_table_generator(
+          product_lookup_table_2_bits_encrypted_mul, 2, 2);
 
       int set_degree_ok = shortint_ciphertext_set_degree(ct_right, degree_to_set);
       assert(set_degree_ok == 0);
@@ -159,14 +153,14 @@ void test_shortint_bivariate_pbs_2_bits_message(void) {
       int decrypt_non_assign_ok = shortint_client_key_decrypt(cks, ct_out, &result_non_assign);
       assert(decrypt_non_assign_ok == 0);
 
-      assert(result_non_assign == product_accumulator_2_bits_encrypted_mul(left_val, right_val));
+      assert(result_non_assign == product_lookup_table_2_bits_encrypted_mul(left_val, right_val));
 
       int pbs_assign_ok = shortint_server_key_bivariate_programmable_bootstrap_assign(
-          sks, accumulator, ct_out, ct_right);
+          sks, lookup_table, ct_out, ct_right);
       assert(pbs_assign_ok == 0);
 
       degree_to_set =
-          (size_t)get_max_value_of_accumulator_generator(double_accumulator_2_bits_message, 2);
+          (size_t)get_max_value_of_lookup_table_generator(double_lookup_table_2_bits_message, 2);
 
       set_degree_ok = shortint_ciphertext_set_degree(ct_out, degree_to_set);
       assert(set_degree_ok == 0);
@@ -176,18 +170,17 @@ void test_shortint_bivariate_pbs_2_bits_message(void) {
       assert(decrypt_assign_ok == 0);
 
       assert(result_assign ==
-             product_accumulator_2_bits_encrypted_mul(result_non_assign, right_val));
+             product_lookup_table_2_bits_encrypted_mul(result_non_assign, right_val));
 
-      destroy_shortint_ciphertext(ct_left);
-      destroy_shortint_ciphertext(ct_right);
-      destroy_shortint_ciphertext(ct_out);
+      shortint_destroy_ciphertext(ct_left);
+      shortint_destroy_ciphertext(ct_right);
+      shortint_destroy_ciphertext(ct_out);
     }
   }
 
-  destroy_shortint_bivariate_pbs_accumulator(accumulator);
-  destroy_shortint_client_key(cks);
-  destroy_shortint_server_key(sks);
-  destroy_shortint_parameters(params);
+  shortint_destroy_bivariate_pbs_lookup_table(lookup_table);
+  shortint_destroy_client_key(cks);
+  shortint_destroy_server_key(sks);
 }
 
 int main(void) {

tfhe/cbindgen.toml

@@ -10,7 +10,7 @@ language = "C"
 
 ############## Options for Wrapping the Contents of the Header #################
 
-header = "// Copyright © 2022 ZAMA.\n// All rights reserved."
+header = "// Copyright © 2023 ZAMA.\n// All rights reserved."
 # trailer = "/* Text to put at the end of the generated file */"
 include_guard = "TFHE_RS_C_API_H"
 # pragma_once = true
@@ -107,7 +107,6 @@ allow_static_const = true
 allow_constexpr = false
 sort_by = "Name"
 
-
 [macro_expansion]
 bitflags = false
 

tfhe/docs/Boolean/parameters.md

@@ -1,44 +0,0 @@
-# Cryptographic Parameters
-
-## Default parameters
-
-The TFHE cryptographic scheme relies on a variant of [Regev cryptosystem](https://cims.nyu.edu/\~regev/papers/lwesurvey.pdf), and is based on a problem so hard to solve that it is even post-quantum resistant.
-
-In practice, you need to tune some cryptographic parameters in order to ensure both the correctness of the result and the security of the computation.
-
-To make it simpler, **we provide two sets of parameters**, which ensure correct computations for a certain probability with the standard security of 128 bits. There exists an error probability due to the probabilistic nature of the encryption, which requires adding randomness (called noise) following a Gaussian distribution. If this noise is too large, the decryption will not give a correct result. There is a trade-off between efficiency and correctness: generally, using a less efficient parameter set (in terms of computation time) leads to a smaller risk of having an error during homomorphic evaluation.
-
-In the two proposed sets of parameters, the only difference lies in this probability error. The default parameter set ensures a probability error of at most $$2^{-40}$$ when computing a programmable bootstrapping (i.e., any gates but the `not`). The other one is closer to the error probability claimed into the original [TFHE paper](https://eprint.iacr.org/2018/421), namely $$2^{-165}$$, but it is up-to-date regarding security requirements.
-
-The following array summarizes this:
-
-|     Parameter set     | Error probability |
-| :-------------------: | :---------------: |
-|  DEFAULT\_PARAMETERS  |    $$2^{-40}$$    |
-| TFHE\_LIB\_PARAMETERS |    $$2^{-165}$$   |
-
-## User-defined parameters
-
-Note that, if you desire, you can also create your own set of parameters. This is an `unsafe` operation as failing to properly fix the parameters will potentially result in an incorrect and/or insecure computation:
-
-```rust
-use tfhe::boolean::prelude::*;
-
-fn main() {
-// WARNING: might be insecure and/or incorrect
-// You can create your own set of parameters
-    let parameters = unsafe {
-        BooleanParameters::new(
-            LweDimension(586),
-            GlweDimension(2),
-            PolynomialSize(512),
-            StandardDev(0.00008976167396834998),
-            StandardDev(0.00000002989040792967434),
-            DecompositionBaseLog(8),
-            DecompositionLevelCount(2),
-            DecompositionBaseLog(2),
-            DecompositionLevelCount(5),
-        )
-    };
-}
-```

tfhe/docs/README.md

@@ -1,12 +1,12 @@
 # What is TFHE-rs?
 
-<mark style="background-color:yellow;">⭐️</mark> [<mark style="background-color:yellow;">Star the repo on Github</mark>](https://github.com/zama-ai/tfhe-rs) <mark style="background-color:yellow;">| 🗣</mark> [<mark style="background-color:yellow;">Community support forum</mark> ](https://community.zama.ai)<mark style="background-color:yellow;">| 📁</mark> [<mark style="background-color:yellow;">Contribute to the project</mark>](https://docs.zama.ai/tfhe-rs/developers/contributing)
+📁 [Github](https://github.com/zama-ai/tfhe-rs) | 💛 [Community support](https://zama.ai/community) | 🟨 [Zama Bounty Program](https://github.com/zama-ai/bounty-program)
 
 ![](\_static/tfhe-rs-doc-home.png)
 
-TFHE-rs is a pure Rust implementation of TFHE for boolean and small integer arithmetics over encrypted data. It includes a Rust and C API, as well as a client-side WASM API.
+TFHE-rs is a pure Rust implementation of TFHE for Boolean and integer arithmetics over encrypted data. It includes a Rust and C API, as well as a client-side WASM API.
 
-TFHE-rs is meant for developers and researchers who want full control over what they can do with TFHE, while not having to worry about the low level implementation.
+TFHE-rs is meant for developers and researchers who want full control over what they can do with TFHE, while not worrying about the low level implementation.
 
 The goal is to have a stable, simple, high-performance, and production-ready library for all the advanced features of TFHE.
 

tfhe/docs/SUMMARY.md

@@ -3,32 +3,50 @@
 * [What is TFHE-rs?](README.md)
 
 ## Getting Started
+* [Installation](getting_started/installation.md)
+* [Quick Start](getting_started/quick_start.md)
+* [Operations](getting_started/operations.md)
+* [Benchmarks](getting_started/benchmarks.md)
+* [Security and Cryptography](getting_started/security_and_cryptography.md)
 
-* [Installation](getting\_started/installation.md)
-* [Quick Start](getting\_started/quick\_start.md)
-* [Supported Operations](getting\_started/operations.md)
-* [Benchmarks](getting\_started/benchmarks.md)
-* [Security and Cryptography](getting\_started/security\_and\_cryptography.md)
+## Tutorials
+* [Homomorphic Parity Bit](tutorials/parity_bit.md)
+* [Homomorphic Case Changing on Latin String](tutorials/latin_fhe_string.md)
 
-## Boolean
-* [Tutorial](Boolean/tutorial.md)
-* [Operations](Boolean/operations.md)
-* [Cryptographic Parameters](Boolean/parameters.md)
-* [Serialization/Deserialization](Boolean/serialization.md)
+## How To
+* [Configure Rust](how_to/rust_configuration.md)
+* [Serialize/Deserialize](how_to/serialization.md)
+* [Compress Ciphertexts/Keys](how_to/compress.md)
+* [Use Public Key Encryption](how_to/public_key.md)
+* [Use Trivial Ciphertext](how_to/trivial_ciphertext.md)
+* [Use Parallelized PBS](how_to/parallelized_pbs.md)
+* [Use the C API](how_to/c_api.md)
+* [Use the JS on WASM API](how_to/js_on_wasm_api.md)
 
-## Shortint
-* [Tutorial](shortint/tutorial.md)
-* [Operations](shortint/operations.md)
-* [Cryptographic Parameters](shortint/parameters.md)
-* [Serialization/Deserialization](shortint/serialization.md)
+## Fine-grained APIs
+* [Quick Start](fine_grained_api/quick_start.md)
+* [Boolean](fine_grained_api/Boolean/readme.md)
+    * [Operations](fine_grained_api/Boolean/operations.md)
+    * [Cryptographic Parameters](fine_grained_api/Boolean/parameters.md)
+    * [Serialization/Deserialization](fine_grained_api/Boolean/serialization.md)
 
-## C API
-* [Tutorial](c_api/tutorial.md)
+* [Shortint](fine_grained_api/shortint/readme.md)
+    * [Operations](fine_grained_api/shortint/operations.md)
+    * [Cryptographic Parameters](fine_grained_api/shortint/parameters.md)
+    * [Serialization/Deserialization](fine_grained_api/shortint/serialization.md)
 
-## JS on WASM API
-* [Tutorial](js_on_wasm_api/tutorial.md)
+* [Integer](fine_grained_api/integer/readme.md)
+    * [Operations](fine_grained_api/integer/operations.md)
+    * [Cryptographic Parameters](fine_grained_api/integer/parameters.md)
+    * [Serialization/Deserialization](fine_grained_api/integer/serialization.md)
 
-## Low-Level Core Cryptography
+## Application Tutorials
+* [SHA256 with *Boolean API*](application_tutorials/sha256_bool.md)
+* [Dark Market with *Integer API*](application_tutorials/dark_market.md)
+* [Homomorphic Regular Expressions *Integer API*](application_tutorials/regex.md)
+
+
+## Crypto Core API [Advanced users]
 * [Quick Start](core_crypto/presentation.md)
 * [Tutorial](core_crypto/tutorial.md)
 

tfhe/docs/application_tutorials/dark_market.md

@@ -0,0 +1,479 @@
+# Dark Market Tutorial
+
+In this tutorial, we are going to build a dark market application using TFHE-rs. A dark market is a marketplace where
+buy and sell orders are not visible to the public before they are filled. Different algorithms aim to
+solve this problem, we are going to implement the algorithm defined [in this paper](https://eprint.iacr.org/2022/923.pdf) with TFHE-rs.
+
+We will first implement the algorithm in plain Rust and then we will see how to use TFHE-rs to
+implement the same algorithm with FHE.
+
+In addition, we will also implement a modified version of the algorithm that allows for more concurrent operations which
+improves the performance in hardware where there are multiple cores.
+
+## Specifications
+
+#### Inputs:
+
+* A list of sell orders where each sell order is only defined in volume terms, it is assumed that the price is fetched
+  from a different source.
+* A list of buy orders where each buy order is only defined in volume terms, it is assumed that the price is fetched
+  from a different source.
+
+#### Input constraints:
+
+* The sell and buy orders are within the range [1,100].
+* The maximum number of sell and buy orders is 500, respectively.
+
+#### Outputs:
+
+There is no output returned at the end of the algorithm. Instead, the algorithm makes changes on the given input lists.
+The number of filled orders is written over the original order count in the respective lists. If it is not possible to
+fill the orders, the order count is set to zero.
+
+#### Example input and output:
+
+##### Example 1:
+
+|        | Sell               | Buy       |
+|--------|--------------------|-----------|
+| Input  | [ 5, 12, 7, 4, 3 ] | [ 19, 2 ] |
+| Output | [ 5, 12, 4, 0, 0 ] | [ 19, 2 ] |
+
+Last three indices of the filled sell orders are zero because there is no buy orders to match them.
+
+##### Example 2:
+
+|        | Sell              | Buy                  |
+|--------|-------------------|----------------------|
+| Input  | [ 3, 1, 1, 4, 2 ] | [ 5, 3, 3, 2, 4, 1 ] |
+| Output | [ 3, 1, 1, 4, 2 ] | [ 5, 3, 3, 0, 0, 0 ] |
+
+Last three indices of the filled buy orders are zero because there is no sell orders to match them.
+
+## Plain Implementation
+
+1. Calculate the total sell volume and the total buy volume.
+
+```rust
+let total_sell_volume: u16 = sell_orders.iter().sum();
+let total_buy_volume: u16 = buy_orders.iter().sum();
+```
+
+2. Find the total volume that will be transacted. In the paper, this amount is calculated with the formula:
+
+```
+(total_sell_volume > total_buy_volume) * (total_buy_volume − total_sell_volume) + total_sell_volume
+```
+
+When closely observed, we can see that this formula can be replaced with the `min` function. Therefore, we calculate this
+value by taking the minimum of the total sell volume and the total buy volume.
+
+```rust
+let total_volume = std::cmp::min(total_buy_volume, total_sell_volume);
+```
+
+3. Beginning with the first item, start filling the sell orders one by one. We apply the `min` function replacement also
+   here.
+
+```rust
+let mut volume_left_to_transact = total_volume;
+for sell_order in sell_orders.iter_mut() {
+    let filled_amount = std::cmp::min(volume_left_to_transact, *sell_order);
+    *sell_order = filled_amount;
+    volume_left_to_transact -= filled_amount;
+}
+```
+
+The number of orders that are filled is indicated by modifying the input list. For example, if the first sell order is
+1000 and the total volume is 500, then the first sell order will be modified to 500 and the second sell order will be
+modified to 0.
+
+4. Do the fill operation also for the buy orders.
+
+```rust
+let mut volume_left_to_transact = total_volume;
+for buy_order in buy_orders.iter_mut() {
+    let filled_amount = std::cmp::min(volume_left_to_transact, *buy_order);
+    *buy_order = filled_amount;
+    volume_left_to_transact -= filled_amount;
+}
+```
+
+#### The complete algorithm in plain Rust:
+
+```rust
+fn volume_match_plain(sell_orders: &mut Vec<u16>, buy_orders: &mut Vec<u16>) {
+    let total_sell_volume: u16 = sell_orders.iter().sum();
+    let total_buy_volume: u16 = buy_orders.iter().sum();
+
+    let total_volume = std::cmp::min(total_buy_volume, total_sell_volume);
+
+    let mut volume_left_to_transact = total_volume;
+    for sell_order in sell_orders.iter_mut() {
+        let filled_amount = std::cmp::min(volume_left_to_transact, *sell_order);
+        *sell_order = filled_amount;
+        volume_left_to_transact -= filled_amount;
+    }
+
+    let mut volume_left_to_transact = total_volume;
+    for buy_order in buy_orders.iter_mut() {
+        let filled_amount = std::cmp::min(volume_left_to_transact, *buy_order);
+        *buy_order = filled_amount;
+        volume_left_to_transact -= filled_amount;
+    }
+}
+```
+
+## FHE Implementation
+
+For the FHE implementation, we first start with finding the right bit size for our algorithm to work without
+overflows.
+
+The variables that are declared in the algorithm and their maximum values are described in the table below:
+
+| Variable                | Maximum Value | Bit Size |
+|-------------------------|---------------|----------|
+| total_sell_volume       | 50000         | 16       |
+| total_buy_volume        | 50000         | 16       |
+| total_volume            | 50000         | 16       |
+| volume_left_to_transact | 50000         | 16       |
+| sell_order              | 100           | 7        |
+| buy_order               | 100           | 7        |
+
+As we can observe from the table, we need **16 bits of message space** to be able to run the algorithm without
+overflows. TFHE-rs provides different presets for the different bit sizes. Since we need 16 bits of message, we are
+going to use the `integer` module to implement the algorithm.
+
+Here are the input types of our algorithm:
+
+* `sell_orders` is of type `Vec<tfhe::integer::RadixCipherText>`
+* `buy_orders` is of type `Vec<tfhe::integer::RadixCipherText>`
+* `server_key` is of type `tfhe::integer::ServerKey`
+
+Now, we can start implementing the algorithm with FHE:
+
+1. Calculate the total sell volume and the total buy volume.
+
+```rust
+let mut total_sell_volume = server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS);
+for sell_order in sell_orders.iter_mut() {
+    server_key.smart_add_assign(&mut total_sell_volume, sell_order);
+}
+
+let mut total_buy_volume = server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS);
+for buy_order in buy_orders.iter_mut() {
+    server_key.smart_add_assign(&mut total_buy_volume, buy_order);
+}
+```
+
+2. Find the total volume that will be transacted by taking the minimum of the total sell volume and the total buy
+   volume.
+
+```rust
+let total_volume = server_key.smart_min(&mut total_sell_volume, &mut total_buy_volume);
+```
+
+3. Beginning with the first item, start filling the sell and buy orders one by one. We can create `fill_orders` closure to 
+reduce code duplication since the code for filling buy orders and sell orders are the same.
+
+```rust
+let fill_orders = |orders: &mut [RadixCiphertext]| {
+    let mut volume_left_to_transact = total_volume.clone();
+    for mut order in orders.iter_mut() {
+        let mut filled_amount = server_key.smart_min(&mut volume_left_to_transact, &mut order);
+        server_key.smart_sub_assign(&mut volume_left_to_transact, &mut filled_amount);
+        *order = filled_amount;
+    }
+};
+
+fill_orders(sell_orders);
+fill_orders(buy_orders);
+```
+
+#### The complete algorithm in TFHE-rs:
+
+```rust
+const NUMBER_OF_BLOCKS: usize = 8;
+
+fn volume_match_fhe(
+    sell_orders: &mut [RadixCiphertext],
+    buy_orders: &mut [RadixCiphertext],
+    server_key: &ServerKey,
+) {
+    let mut total_sell_volume = server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS);
+    for sell_order in sell_orders.iter_mut() {
+        server_key.smart_add_assign(&mut total_sell_volume, sell_order);
+    }
+
+    let mut total_buy_volume = server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS);
+    for buy_order in buy_orders.iter_mut() {
+        server_key.smart_add_assign(&mut total_buy_volume, buy_order);
+    }
+
+    let total_volume = server_key.smart_min(&mut total_sell_volume, &mut total_buy_volume);
+
+    let fill_orders = |orders: &mut [RadixCiphertext]| {
+        let mut volume_left_to_transact = total_volume.clone();
+        for mut order in orders.iter_mut() {
+            let mut filled_amount = server_key.smart_min(&mut volume_left_to_transact, &mut order);
+            server_key.smart_sub_assign(&mut volume_left_to_transact, &mut filled_amount);
+            *order = filled_amount;
+        }
+    };
+
+    fill_orders(sell_orders);
+    fill_orders(buy_orders);
+}
+
+```
+
+### Optimizing the implementation
+
+* TFHE-rs provides parallelized implementations of the operations. We can use these parallelized
+  implementations to speed up the algorithm. For example, we can use `smart_add_assign_parallelized` instead of 
+  `smart_add_assign`.
+
+* We can parallelize vector sum with Rayon and `reduce` operation.
+```rust
+let parallel_vector_sum = |vec: &mut [RadixCiphertext]| {
+    vec.to_vec().into_par_iter().reduce(
+        || server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS),
+        |mut acc: RadixCiphertext, mut ele: RadixCiphertext| { 
+            server_key.smart_add_parallelized(&mut acc, &mut ele)
+        },
+    )
+};
+```
+
+* We can run vector summation on `buy_orders` and `sell_orders` in parallel since these operations do not depend on each other.
+```rust
+let (mut total_sell_volume, mut total_buy_volume) =
+    rayon::join(|| vector_sum(sell_orders), || vector_sum(buy_orders));
+```
+
+* We can match sell and buy orders in parallel since the matching does not depend on each other.
+```rust
+rayon::join(|| fill_orders(sell_orders), || fill_orders(buy_orders));
+```
+
+#### Optimized algorithm
+```rust
+fn volume_match_fhe_parallelized(
+    sell_orders: &mut [RadixCiphertext],
+    buy_orders: &mut [RadixCiphertext],
+    server_key: &ServerKey,
+) {
+    let parallel_vector_sum = |vec: &mut [RadixCiphertext]| {
+        vec.to_vec().into_par_iter().reduce(
+            || server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS),
+            |mut acc: RadixCiphertext, mut ele: RadixCiphertext| {
+                server_key.smart_add_parallelized(&mut acc, &mut ele)
+            },
+        )
+    };
+
+    let (mut total_sell_volume, mut total_buy_volume) = rayon::join(
+        || parallel_vector_sum(sell_orders),
+        || parallel_vector_sum(buy_orders),
+    );
+
+    let total_volume =
+        server_key.smart_min_parallelized(&mut total_sell_volume, &mut total_buy_volume);
+
+    let fill_orders = |orders: &mut [RadixCiphertext]| {
+        let mut volume_left_to_transact = total_volume.clone();
+        for mut order in orders.iter_mut() {
+            let mut filled_amount =
+                server_key.smart_min_parallelized(&mut volume_left_to_transact, &mut order);
+            server_key
+                .smart_sub_assign_parallelized(&mut volume_left_to_transact, &mut filled_amount);
+            *order = filled_amount;
+        }
+    };
+    
+    rayon::join(|| fill_orders(sell_orders), || fill_orders(buy_orders));
+}
+```
+
+## Modified Algorithm
+
+When observed closely, there is only a small amount of concurrency introduced in the `fill_orders` part of the algorithm.
+The reason is that the `volume_left_to_transact` is shared between all the orders and should be modified sequentially. 
+This means that the orders cannot be filled in parallel. If we can somehow remove this dependency, we can fill the orders in parallel.
+
+In order to do so, we closely observe the function of `volume_left_to_transact` variable in the algorithm. We can see that it is being used to check whether we can fill the current order or not.
+Instead of subtracting the current order value from `volume_left_to_transact` in each loop, we can add this value to the next order
+index and check the availability by comparing the current order value with the total volume. If the current order value
+(now representing the sum of values before this order plus this order) is smaller than the total number of matching orders,
+we can safely fill all the orders and continue the loop. If not, we should partially fill the orders with what is left from
+matching orders. 
+
+We will call the new list the "prefix sum" of the array. 
+
+The new version for the plain `fill_orders` is as follows:
+```rust
+let fill_orders = |orders: &mut [u64], prefix_sum: &[u64], total_orders: u64|{
+    orders.iter().for_each(|order : &mut u64| {
+        if (total_orders >= prefix_sum[i]) {
+            continue;
+        } else if total_orders >= prefix_sum.get(i-1).unwrap_or(0) {
+            *order = total_orders - prefix_sum.get(i-1).unwrap_or(0);
+        } else {
+            *order = 0;
+        }
+    });
+};
+```
+
+To write this new function we need transform the conditional code into a mathematical expression since FHE does not support conditional operations.
+```rust
+
+let fill_orders = |orders: &mut [u64], prefix_sum: &[u64], total_orders: u64| {
+    orders.iter().for_each(|order| : &mut){
+        *order = *order + ((total_orders >= prefix_sum - std::cmp::min(total_orders, prefix_sum.get(i - 1).unwrap_or(&0).clone()) - *order);
+    }
+};
+```
+
+New `fill_order` function requires a prefix sum array. We are going to calculate this prefix sum array in parallel 
+with the algorithm described [here](https://developer.nvidia.com/gpugems/gpugems3/part-vi-gpu-computing/chapter-39-parallel-prefix-sum-scan-cuda).
+
+The sample code in the paper is written in CUDA. When we try to implement the algorithm in Rust we see that the compiler does not allow us to do so.
+The reason for that is while the algorithm does not access the same array element in any of the threads(the index calculations using `d` and `k` values never overlap), 
+Rust compiler cannot understand this and does not let us share the same array between threads. 
+So we modify how the algorithm is implemented, but we don't change the algorithm itself.
+
+Here is the modified version of the algorithm in TFHE-rs:
+```rust
+fn volume_match_fhe_modified(
+    sell_orders: &mut [RadixCiphertext],
+    buy_orders: &mut [RadixCiphertext],
+    server_key: &ServerKey,
+) {
+    let compute_prefix_sum = |arr: &[RadixCiphertext]| {
+        if arr.is_empty() {
+            return arr.to_vec();
+        }
+        let mut prefix_sum: Vec<RadixCiphertext> = (0..arr.len().next_power_of_two())
+            .into_par_iter()
+            .map(|i| {
+                if i < arr.len() {
+                    arr[i].clone()
+                } else {
+                    server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS)
+                }
+            })
+            .collect();
+        // Up sweep
+        for d in 0..(prefix_sum.len().ilog2() as u32) {
+            prefix_sum
+                .par_chunks_exact_mut(2_usize.pow(d + 1))
+                .for_each(move |chunk| {
+                    let length = chunk.len();
+                    let mut left = chunk.get((length - 1) / 2).unwrap().clone();
+                    server_key.smart_add_assign_parallelized(chunk.last_mut().unwrap(), &mut left)
+                });
+        }
+        // Down sweep
+        let last = prefix_sum.last().unwrap().clone();
+        *prefix_sum.last_mut().unwrap() = server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS);
+        for d in (0..(prefix_sum.len().ilog2() as u32)).rev() {
+            prefix_sum
+                .par_chunks_exact_mut(2_usize.pow(d + 1))
+                .for_each(move |chunk| {
+                    let length = chunk.len();
+                    let t = chunk.last().unwrap().clone();
+                    let mut left = chunk.get((length - 1) / 2).unwrap().clone();
+                    server_key.smart_add_assign_parallelized(chunk.last_mut().unwrap(), &mut left);
+                    chunk[(length - 1) / 2] = t;
+                });
+        }
+        prefix_sum.push(last);
+        prefix_sum[1..=arr.len()].to_vec()
+    };
+
+    println!("Creating prefix sum arrays...");
+    let time = Instant::now();
+    let (prefix_sum_sell_orders, prefix_sum_buy_orders) = rayon::join(
+        || compute_prefix_sum(sell_orders),
+        || compute_prefix_sum(buy_orders),
+    );
+    println!("Created prefix sum arrays in {:?}", time.elapsed());
+
+    let fill_orders = |total_orders: &RadixCiphertext,
+                        orders: &mut [RadixCiphertext],
+                        prefix_sum_arr: &[RadixCiphertext]| {
+        orders
+            .into_par_iter()
+            .enumerate()
+            .for_each(move |(i, order)| {
+                server_key.smart_add_assign_parallelized(
+                    order,
+                    &mut server_key.smart_mul_parallelized(
+                        &mut server_key
+                            .smart_ge_parallelized(&mut order.clone(), &mut total_orders.clone()),
+                        &mut server_key.smart_sub_parallelized(
+                            &mut server_key.smart_sub_parallelized(
+                                &mut total_orders.clone(),
+                                &mut server_key.smart_min_parallelized(
+                                    &mut total_orders.clone(),
+                                    &mut prefix_sum_arr
+                                        .get(i - 1)
+                                        .unwrap_or(
+                                            &server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS),
+                                        )
+                                        .clone(),
+                                ),
+                            ),
+                            &mut order.clone(),
+                        ),
+                    ),
+                );
+            });
+    };
+
+    let total_buy_orders = &mut prefix_sum_buy_orders
+        .last()
+        .unwrap_or(&server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS))
+        .clone();
+
+    let total_sell_orders = &mut prefix_sum_sell_orders
+        .last()
+        .unwrap_or(&server_key.create_trivial_zero_radix(NUMBER_OF_BLOCKS))
+        .clone();
+
+    println!("Matching orders...");
+    let time = Instant::now();
+    rayon::join(
+        || fill_orders(total_sell_orders, buy_orders, &prefix_sum_buy_orders),
+        || fill_orders(total_buy_orders, sell_orders, &prefix_sum_sell_orders),
+    );
+    println!("Matched orders in {:?}", time.elapsed());
+}
+```
+
+## Running the tutorial
+
+The plain, FHE and parallel FHE implementations can be run by providing respective arguments as described below.
+
+```bash
+# Runs FHE implementation
+cargo run --release --package tfhe --example dark_market --features="integer internal-keycache" -- fhe
+
+# Runs parallelized FHE implementation
+cargo run --release --package tfhe --example dark_market --features="integer internal-keycache" -- fhe-parallel
+
+# Runs modified FHE implementation
+cargo run --release --package tfhe --example dark_market --features="integer internal-keycache" -- fhe-modified
+
+# Runs plain implementation
+cargo run --release --package tfhe --example dark_market --features="integer internal-keycache" -- plain
+
+# Multiple implementations can be run within same instance
+cargo run --release --package tfhe --example dark_market --features="integer internal-keycache" -- plain fhe-parallel
+```
+
+## Conclusion
+
+In this tutorial, we've learned how to implement the volume matching algorithm described [in this paper](https://eprint.iacr.org/2022/923.pdf) in plain Rust and in TFHE-rs. 
+We've identified the right bit size for our problem at hand, used operations defined in `TFHE-rs`, and introduced concurrency to the algorithm to increase its performance.