Use intersphinx links for deep learning (#5859)

* Use intersphinx links for deep learning

* Update deep-learning-rocm.rst

remove Taichi

* Update deep-learning-rocm.rst

Change Install link to "link"

* Apply suggestion from @randyh62

OK

.azuredevops/components/HIP.yml

@@ -34,7 +34,6 @@ parameters:
   default:
     - cmake
     - libnuma-dev
-    - libsimde-dev
     - mesa-common-dev
     - ninja-build
     - ocl-icd-libopencl1

.azuredevops/components/origami.yml

@@ -39,7 +39,6 @@ parameters:
     - python3
     - python3-dev
     - python3-pip
-    - python3-venv
     - libgtest-dev
     - libboost-filesystem-dev
     - libboost-program-options-dev
@@ -47,8 +46,6 @@ parameters:
   type: object
   default:
     - nanobind>=2.0.0
-    - pytest
-    - pytest-cov
 - name: rocmDependencies
   type: object
   default:
@@ -75,10 +72,8 @@ parameters:
       - { os: ubuntu2204, packageManager: apt }
       - { os: almalinux8, packageManager: dnf }
     testJobs:
+      - { os: ubuntu2204, packageManager: apt, target: gfx942 }
       - { os: ubuntu2204, packageManager: apt, target: gfx90a }
-      # - { os: ubuntu2204, packageManager: apt, target: gfx1100 }
-      # - { os: ubuntu2204, packageManager: apt, target: gfx1151 }
-      # - { os: ubuntu2204, packageManager: apt, target: gfx1201 }
 - name: downstreamComponentMatrix
   type: object
   default:
@@ -121,11 +116,6 @@ jobs:
       parameters:
         dependencyList:
           - gtest
-    - ${{ if ne(job.os, 'almalinux8') }}:
-      - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-vendor.yml
-        parameters:
-          dependencyList:
-            - catch2
     - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/checkout.yml
       parameters:
         checkoutRepo: ${{ parameters.checkoutRepo }}
@@ -147,7 +137,6 @@ jobs:
           -DORIGAMI_BUILD_SHARED_LIBS=ON
           -DORIGAMI_ENABLE_PYTHON=ON
           -DORIGAMI_BUILD_TESTING=ON
-          -DORIGAMI_ENABLE_FETCH=ON
           -GNinja
     - ${{ if ne(job.os, 'almalinux8') }}:
       - task: PublishPipelineArtifact@1
@@ -180,6 +169,7 @@ jobs:
       dependsOn: origami_build_${{ job.os }}
       condition:
         and(succeeded(),
+          eq(variables['ENABLE_${{ upper(job.target) }}_TESTS'], 'true'),
           not(containsValue(split(variables['DISABLED_${{ upper(job.target) }}_TESTS'], ','), '${{ parameters.componentName }}')),
           eq(${{ parameters.aggregatePipeline }}, False)
         )
@@ -190,30 +180,30 @@ jobs:
       workspace:
         clean: all
       steps:
+      - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/checkout.yml
+        parameters:
+          checkoutRepo: ${{ parameters.checkoutRepo }}
+          sparseCheckoutDir: ${{ parameters.sparseCheckoutDir }}
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-other.yml
         parameters:
           aptPackages: ${{ parameters.aptPackages }}
           pipModules: ${{ parameters.pipModules }}
           packageManager: ${{ job.packageManager }}
-      - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-cmake-custom.yml
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/preamble.yml
-      - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/checkout.yml
-        parameters:
-          checkoutRepo: ${{ parameters.checkoutRepo }}
-          sparseCheckoutDir: ${{ parameters.sparseCheckoutDir }}
-      - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-vendor.yml
-        parameters:
-          dependencyList:
-            - gtest
-      - ${{ if ne(job.os, 'almalinux8') }}:
-        - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-vendor.yml
-          parameters:
-            dependencyList:
-              - catch2
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/local-artifact-download.yml
         parameters:
           preTargetFilter: ${{ parameters.componentName }}
           os: ${{ job.os }}
+      - task: DownloadPipelineArtifact@2
+        displayName: 'Download Build Directory Artifact'
+        inputs:
+          artifact: '${{ parameters.componentName }}_${{ job.os }}_build_dir'
+          path: '$(Agent.BuildDirectory)/s/build'
+      - task: DownloadPipelineArtifact@2
+        displayName: 'Download Python Source Artifact'
+        inputs:
+          artifact: '${{ parameters.componentName }}_${{ job.os }}_python_src'
+          path: '$(Agent.BuildDirectory)/s/python'
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/dependencies-rocm.yml
         parameters:
           checkoutRef: ${{ parameters.checkoutRef }}
@@ -222,72 +212,25 @@ jobs:
           gpuTarget: ${{ job.target }}
           ${{ if parameters.triggerDownstreamJobs }}:
             downstreamAggregateNames: ${{ parameters.downstreamAggregateNames }}
-      - task: CMake@1
-        displayName: 'Origami Test CMake Configuration'
-        inputs:
-          cmakeArgs: >-
-            -DCMAKE_PREFIX_PATH=$(Agent.BuildDirectory)/rocm;$(Agent.BuildDirectory)/vendor
-            -DCMAKE_CXX_COMPILER=$(Agent.BuildDirectory)/rocm/llvm/bin/amdclang++
-            -DORIGAMI_BUILD_SHARED_LIBS=ON
-            -DORIGAMI_ENABLE_PYTHON=ON
-            -DORIGAMI_BUILD_TESTING=ON
-            -GNinja
-            $(Agent.BuildDirectory)/s
-      - task: Bash@3
-        displayName: 'Build Origami Tests and Python Bindings'
-        inputs:
-          targetType: inline
-          workingDirectory: build
-          script: |
-            cmake --build . --target origami-tests origami_python -- -j$(nproc)
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/gpu-diagnostics.yml
-      # Run tests using CTest (discovers and runs both C++ and Python tests)
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/test.yml
         parameters:
           componentName: ${{ parameters.componentName }}
           os: ${{ job.os }}
-          testDir: 'build'
+          testDir: '$(Agent.BuildDirectory)/rocm/bin'
-          testParameters: '--output-on-failure --force-new-ctest-process --output-junit test_output.xml'
+          testExecutable: './origami-tests'
-      # Test pip install workflow
+          testParameters: '--yaml origami-tests.yaml --gtest_output=xml:./test_output.xml --gtest_color=yes'
-      # - task: Bash@3
+      - script: |
-      #   displayName: 'Test Pip Install'
+          set -e
-      #   inputs:
+          export PYTHONPATH=$(Agent.BuildDirectory)/s/build/python:$PYTHONPATH
-      #     targetType: inline
-      #     script: |
-      #       set -e
 
-      #       echo "==================================================================="
+          echo "--- Running origami_test.py ---"
-      #       echo "Testing pip install workflow (pip install -e .)"
+          python3 $(Agent.BuildDirectory)/s/python/origami_test.py
-      #       echo "==================================================================="
           
-      #       # Set environment variables for pip install CMake build
+          echo "--- Running origami_grid_test.py ---"
-      #       export ROCM_PATH=$(Agent.BuildDirectory)/rocm
+          python3 $(Agent.BuildDirectory)/s/python/origami_grid_test.py
-      #       export CMAKE_PREFIX_PATH=$(Agent.BuildDirectory)/rocm:$(Agent.BuildDirectory)/vendor
+        displayName: 'Run Python Binding Tests'
-      #       export CMAKE_CXX_COMPILER=$(Agent.BuildDirectory)/rocm/llvm/bin/amdclang++
+        condition: succeeded()
-            
-      #       echo "ROCM_PATH: $ROCM_PATH"
-      #       echo "CMAKE_PREFIX_PATH: $CMAKE_PREFIX_PATH"
-      #       echo "CMAKE_CXX_COMPILER: $CMAKE_CXX_COMPILER"
-      #       echo ""
-            
-      #       # Install from source directory
-      #       cd "$(Agent.BuildDirectory)/s/python"
-      #       pip install -e .
-            
-      #       # Verify import works
-      #       echo ""
-      #       echo "Verifying origami can be imported..."
-      #       python3 -c "import origami; print('✓ Successfully imported origami')"
-            
-      #       # Run pytest on installed package
-      #       echo ""
-      #       echo "Running pytest tests..."
-      #       python3 -m pytest tests/ -v -m "not slow" --tb=short
-            
-      #       echo ""
-      #       echo "==================================================================="
-      #       echo "Pip install test completed successfully"
-      #       echo "==================================================================="
       - template: ${{ variables.CI_TEMPLATE_PATH }}/steps/docker-container.yml
         parameters:
           aptPackages: ${{ parameters.aptPackages }}

.azuredevops/components/rocm-examples.yml

@@ -30,7 +30,6 @@ parameters:
     - python3-pip
     - protobuf-compiler
     - libprotoc-dev
-    - libopencv-dev
 - name: pipModules
   type: object
   default:
@@ -65,7 +64,6 @@ parameters:
     - MIVisionX
     - rocm_smi_lib
     - rccl
-    - rocAL
     - rocALUTION
     - rocBLAS
     - rocDecode
@@ -105,7 +103,6 @@ parameters:
     - MIVisionX
     - rocm_smi_lib
     - rccl
-    - rocAL
     - rocALUTION
     - rocBLAS
     - rocDecode

.wordlist.txt

@@ -36,6 +36,7 @@ Andrej
 Arb
 Autocast
 autograd
+Backported
 BARs
 BatchNorm
 BLAS
@@ -203,9 +204,11 @@ GenAI
 GenZ
 GitHub
 Gitpod
+hardcoded
 HBM
 HCA
 HGX
+HLO
 HIPCC
 hipDataType
 HIPExtension
@@ -333,6 +336,7 @@ MoEs
 Mooncake
 Mpops
 Multicore
+multihost
 Multithreaded
 mx
 MXFP
@@ -1027,6 +1031,7 @@ uncacheable
 uncorrectable
 underoptimized
 unhandled
+unfused
 uninstallation
 unmapped
 unsqueeze

RELEASE.md

@@ -270,26 +270,26 @@ The [ROCm examples repository](https://github.com/ROCm/rocm-examples) has been e
 :margin: auto 0 auto auto
 :::{grid}
 :margin: auto 0 auto auto
-* [hipBLASLt](https://rocm.docs.amd.com/projects/hipBLASLt/en/latest/)
+* [hipBLASLt](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/hipBLASLt)
-* [hipSPARSE](https://rocm.docs.amd.com/projects/hipSPARSE/en/latest/)
+* [hipSPARSE](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/hipSPARSE)
-* [hipSPARSELt](https://rocm.docs.amd.com/projects/hipSPARSELt/en/latest/)
+* [hipSPARSELt](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/hipSPARSELt)
-* [hipTensor](https://rocm.docs.amd.com/projects/hipTensor/en/latest/)
+* [hipTensor](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/hipTensor)
 :::
 :::{grid}
 :margin: auto 0 auto auto
-* [rocALUTION](https://rocm.docs.amd.com/projects/rocALUTION/en/latest/)
+* [rocALUTION](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/rocALUTION)
-* [ROCprofiler-SDK](https://rocm.docs.amd.com/projects/rocprofiler-sdk/en/latest/)
+* [ROCprofiler-SDK](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/rocProfiler-SDK)
-* [rocWMMA](https://rocm.docs.amd.com/projects/rocWMMA/en/latest/)
+* [rocWMMA](https://github.com/ROCm/rocm-examples/tree/amd-staging/Libraries/rocWMMA)
 :::
 ::::
 
 Usage examples are now available for the following performance analysis tools:
 
-* [ROCm Compute Profiler](https://rocm.docs.amd.com/projects/rocprofiler-compute/en/latest/index.html)
+* [ROCm Compute Profiler](https://github.com/ROCm/rocm-examples/tree/amd-staging/Tools/rocprof-compute)
-* [ROCm Systems Profiler](https://rocm.docs.amd.com/projects/rocprofiler-systems/en/latest/index.html)
+* [ROCm Systems Profiler](https://github.com/ROCm/rocm-examples/tree/amd-staging/Tools/rocprof-systems)
-* [rocprofv3](https://rocm.docs.amd.com/projects/rocprofiler-sdk/en/latest/how-to/using-rocprofv3.html)
+* [rocprofv3](https://github.com/ROCm/rocm-examples/tree/amd-staging/Tools/rocprofv3)
 
-The complete source code for the [HIP Graph Tutorial](https://rocm.docs.amd.com/projects/HIP/en/latest/tutorial/graph_api.html) is also available as part of the ROCm examples.
+The complete source code for the [HIP Graph Tutorial](https://github.com/ROCm/rocm-examples/tree/amd-staging/HIP-Doc/Tutorials/graph_api) is also available as part of the ROCm examples.
 
 ### ROCm documentation updates
 

docs/compatibility/ml-compatibility/jax-compatibility.rst

@@ -269,6 +269,33 @@ For a complete and up-to-date list of JAX public modules (for example, ``jax.num
   JAX API modules are maintained by the JAX project and is subject to change.
   Refer to the official Jax documentation for the most up-to-date information.
 
+Key features and enhancements for ROCm 7.1
+===============================================================================
+
+- Enabled compilation of multihost HLO runner Python bindings.
+
+  - Backported multihost HLO runner bindings and some related changes to
+    :code:`FunctionalHloRunner`.
+
+  - Added :code:`requirements_lock_3_12` to enable building for Python 3.12.
+
+- Removed hardcoded NHWC convolution layout for ``fp16`` precision to address the performance drops for ``fp16`` precision on gfx12xx GPUs.
+
+
+- ROCprofiler-SDK integration:
+
+  - Integrated ROCprofiler-SDK (v3) to XLA to improve profiling of GPU events,
+    support both time-based and step-based profiling.
+
+  - Added unit tests for :code:`rocm_collector` and :code:`rocm_tracer`.
+
+- Added Triton unsupported conversion from ``f8E4M3FNUZ`` to ``fp16`` with
+  rounding mode.
+
+- Introduced :code:`CudnnFusedConvDecomposer` to revert fused convolutions
+  when :code:`ConvAlgorithmPicker` fails to find a fused algorithm, and removed
+  unfused fallback paths from :code:`RocmFusedConvRunner`.
+
 Key features and enhancements for ROCm 7.0
 ===============================================================================
 

docs/conf.py

@@ -268,6 +268,3 @@ html_context = {
     "granularity_type" : [('Coarse-grained', 'coarse-grained'), ('Fine-grained', 'fine-grained')],
     "scope_type" : [('Device', 'device'), ('System', 'system')]
 }
-
-# Disable figure and table numbering
-numfig = False

docs/how-to/deep-learning-rocm.rst

@@ -24,7 +24,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
       - GitHub
 
     * - :doc:`PyTorch <../compatibility/ml-compatibility/pytorch-compatibility>`
-      - :doc:`Pytorch install <rocm-install-on-linux:install/3rd-party/pytorch-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/pytorch-install>`
       - 
         - Docker image
         - Wheels package
@@ -35,7 +35,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/pytorch"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`TensorFlow <../compatibility/ml-compatibility/tensorflow-compatibility>`
-      - :doc:`TensorFlow install <rocm-install-on-linux:install/3rd-party/tensorflow-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/tensorflow-install>`
       - 
         - Docker image
         - Wheels package
@@ -45,7 +45,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/tensorflow-upstream"><i class="fab fa-github fa-lg"></i></a> 
 
     * - :doc:`JAX <../compatibility/ml-compatibility/jax-compatibility>`
-      - :doc:`JAX install <rocm-install-on-linux:install/3rd-party/jax-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/jax-install>`
       - 
         - Docker image
       - .. raw:: html
@@ -53,7 +53,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/jax"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`verl <../compatibility/ml-compatibility/verl-compatibility>`
-      - :doc:`verl install <rocm-install-on-linux:install/3rd-party/verl-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/verl-install>`
       - 
         - Docker image
       - .. raw:: html
@@ -61,7 +61,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/verl"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`Stanford Megatron-LM <../compatibility/ml-compatibility/stanford-megatron-lm-compatibility>`
-      - :doc:`Stanford Megatron-LM install <rocm-install-on-linux:install/3rd-party/stanford-megatron-lm-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/stanford-megatron-lm-install>`
       - 
         - Docker image
       - .. raw:: html
@@ -69,7 +69,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/Stanford-Megatron-LM"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`DGL <../compatibility/ml-compatibility/dgl-compatibility>`
-      - :doc:`DGL install <rocm-install-on-linux:install/3rd-party/dgl-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/dgl-install>`
       - 
         - Docker image
       - .. raw:: html
@@ -77,24 +77,15 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/dgl"><i class="fab fa-github fa-lg"></i></a> 
 
     * - :doc:`Megablocks <../compatibility/ml-compatibility/megablocks-compatibility>`
-      - :doc:`Megablocks install <rocm-install-on-linux:install/3rd-party/megablocks-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/megablocks-install>`
       - 
         - Docker image
       - .. raw:: html
 
           <a href="https://github.com/ROCm/megablocks"><i class="fab fa-github fa-lg"></i></a>
 
-    * - :doc:`Taichi <../compatibility/ml-compatibility/taichi-compatibility>`
-      - `Taichi install <https://rocm.docs.amd.com/projects/taichi/en/latest/install/taichi-install.html>`__
-      - 
-        - Docker image
-        - Wheels package
-      - .. raw:: html
-
-          <a href="https://github.com/ROCm/taichi"><i class="fab fa-github fa-lg"></i></a>
-
     * - :doc:`Ray <../compatibility/ml-compatibility/ray-compatibility>`
-      - :doc:`Ray install <rocm-install-on-linux:install/3rd-party/ray-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/ray-install>`
       - 
         - Docker image
         - Wheels package
@@ -104,7 +95,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/ray"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`llama.cpp <../compatibility/ml-compatibility/llama-cpp-compatibility>`
-      - :doc:`llama.cpp install <rocm-install-on-linux:install/3rd-party/llama-cpp-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/llama-cpp-install>`
       - 
         - Docker image
         - ROCm Base Docker image
@@ -113,7 +104,7 @@ The table below summarizes information about ROCm-enabled deep learning framewor
           <a href="https://github.com/ROCm/llama.cpp"><i class="fab fa-github fa-lg"></i></a>
 
     * - :doc:`FlashInfer <../compatibility/ml-compatibility/flashinfer-compatibility>`
-      - :doc:`FlashInfer install <rocm-install-on-linux:install/3rd-party/flashinfer-install>`
+      - :doc:`link <rocm-install-on-linux:install/3rd-party/flashinfer-install>`
       - 
         - Docker image
         - ROCm Base Docker image

docs/how-to/rocm-for-ai/fine-tuning/multi-gpu-fine-tuning-and-inference.rst

@@ -44,7 +44,7 @@ Setting up the base implementation environment
 
    .. code-block:: shell
 
-      rocm-smi --showproductname
+      amd-smi static --board
 
 #. Check that your GPUs are available to PyTorch.
 
@@ -65,8 +65,8 @@ Setting up the base implementation environment
 
 .. tip::
 
-   During training and inference, you can check the memory usage by running the ``rocm-smi`` command in your terminal.
+   During training and inference, you can check the memory usage by running the ``amd-smi`` command in your terminal.
-   This tool helps you see shows which GPUs are involved.
+   This tool helps you see which GPUs are involved.
 
 
 .. _fine-tuning-llms-multi-gpu-hugging-face-accelerate:
@@ -130,7 +130,7 @@ After loading the model in this way, the model is fully ready to use the resourc
 torchtune for fine-tuning and inference
 =============================================
 
-`torchtune <https://meta-pytorch.org/torchtune/main/>`_ is a PyTorch-native library for easy single and multi-GPU 
+`torchtune <https://pytorch.org/torchtune/main/>`_ is a PyTorch-native library for easy single and multi-GPU
 model fine-tuning and inference with LLMs.
 
 #. Install torchtune using pip.

docs/how-to/rocm-for-ai/fine-tuning/single-gpu-fine-tuning-and-inference.rst

@@ -44,20 +44,19 @@ Setting up the base implementation environment
 
    .. code-block:: shell
 
-      rocm-smi --showproductname
+      amd-smi static --board
 
    Your output should look like this:
 
    .. code-block:: shell
 
-      ============================ ROCm System Management Interface ============================
+      GPU: 0
-      ====================================== Product Info ======================================
+         BOARD:
-      GPU[0]          : Card Series:          AMD Instinct MI300X OAM
+            MODEL_NUMBER: 102-G39203-0B
-      GPU[0]          : Card model:           0x74a1
+            PRODUCT_SERIAL: PCB079220-1150
-      GPU[0]          : Card vendor:          Advanced Micro Devices, Inc. [AMD/ATI]
+            FRU_ID: 113-AMDG392030B04-100-300000097H
-      GPU[0]          : Card SKU:             MI3SRIOV
+            PRODUCT_NAME: AMD Instinct MI325 OAM
-      ==========================================================================================
+            MANUFACTURER_NAME: AMD
-      ================================== End of ROCm SMI Log ===================================
 
 #. Check that your GPUs are available to PyTorch.
 

docs/how-to/rocm-for-ai/system-setup/prerequisite-system-validation.rst

@@ -31,16 +31,16 @@ in the Instinct documentation for more information.
 Hardware verification with ROCm
 -------------------------------
 
-Use the command ``rocm-smi --setperfdeterminism 1900`` to set the max clock speed up to 1900 MHz
+Use the command ``amd-smi set --perf-determinism 1900`` to set the max clock speed up to 1900 MHz
 instead of the default 2100 MHz. This can reduce the chance of a PCC event lowering the attainable
 GPU clocks. This setting will not be required for new IFWI releases with the production PRC feature.
-You can restore this setting to its default value with the ``rocm-smi -r`` command.
+You can restore this setting to its default value with the ``amd-smi reset --clocks`` command.
 
 Run the command:
 
 .. code-block:: shell
 
-   rocm-smi --setperfdeterminism 1900
+   amd-smi set --perf-determinism 1900
 
 See `Hardware verfication for ROCm <https://instinct.docs.amd.com/projects/amdgpu-docs/en/latest/system-optimization/mi300x.html#hardware-verification-with-rocm>`_
 in the Instinct documentation for more information.

docs/how-to/rocm-for-ai/training/benchmark-docker/previous-versions/megatron-lm-v24.12-dev.rst

@@ -108,16 +108,16 @@ for more information.
 Hardware verification with ROCm
 -------------------------------
 
-Use the command ``rocm-smi --setperfdeterminism 1900`` to set the max clock speed up to 1900 MHz
+Use the command ``amd-smi set --perf-determinism 1900`` to set the max clock speed up to 1900 MHz
 instead of the default 2100 MHz. This can reduce the chance of a PCC event lowering the attainable
 GPU clocks. This setting will not be required for new IFWI releases with the production PRC feature.
-You can restore this setting to its default value with the ``rocm-smi -r`` command.
+You can restore this setting to its default value with the ``amd-smi reset --clocks`` command.
 
 Run the command:
 
 .. code-block:: shell
 
-   rocm-smi --setperfdeterminism 1900
+   amd-smi set --perf-determinism 1900
 
 See `Hardware verification with ROCm <https://instinct.docs.amd.com/projects/amdgpu-docs/en/latest/system-optimization/mi300x.html#hardware-verification-with-rocm>`_ for more information.
 

docs/reference/gpu-arch-specs.rst

@@ -5,7 +5,7 @@
 GPU hardware specifications
 ===========================================
 
-The following tables provide an overview of the hardware specifications for AMD Instinct™ GPUs, and AMD Radeon™ PRO and Radeon™ GPUs.
+The following tables provide an overview of the hardware specifications for AMD Instinct™ GPUs, AMD Radeon™ PRO and Radeon™ GPUs, and AMD Ryzen™ APUs.
 
 For more information about ROCm hardware compatibility, see the ROCm `Compatibility matrix <https://rocm.docs.amd.com/en/latest/compatibility/compatibility-matrix.html>`_.
 
@@ -18,7 +18,7 @@ For more information about ROCm hardware compatibility, see the ROCm `Compatibil
         :name: instinct-arch-spec-table
 
         *
-          - Model
+          - Name
           - Architecture
           - LLVM target name
           - VRAM (GiB)
@@ -297,7 +297,7 @@ For more information about ROCm hardware compatibility, see the ROCm `Compatibil
         :name: radeon-pro-arch-spec-table
 
         *
-          - Model
+          - Name
           - Architecture
           - LLVM target name
 
@@ -539,7 +539,7 @@ For more information about ROCm hardware compatibility, see the ROCm `Compatibil
         :name: radeon-arch-spec-table
 
         *
-          - Model
+          - Name
           - Architecture
           - LLVM target name
           - VRAM (GiB)
@@ -953,6 +953,127 @@ For more information about ROCm hardware compatibility, see the ROCm `Compatibil
           - 9
           - 0
 
+  .. tab-item:: AMD Ryzen APUs
+
+    .. list-table::
+        :header-rows: 1
+        :name: ryzen-arch-spec-table
+
+        *
+          - Name
+          - Graphics model
+          - Architecture
+          - LLVM target name
+          - VRAM (GiB)
+          - Compute Units
+          - Wavefront Size
+          - LDS (KiB)
+          - Infinity Cache (MiB)
+          - L2 Cache (MiB)
+          - Graphics L1 Cache (KiB)
+          - L0 Vector Cache (KiB)
+          - L0 Scalar Cache (KiB)
+          - L0 Instruction Cache (KiB)
+          - VGPR File (KiB)
+          - SGPR File (KiB)
+          - GFXIP Major version
+          - GFXIP Minor version
+        *
+          - AMD Ryzen 7 7840U
+          - Radeon 780M
+          - RDNA3
+          - gfx1103
+          - Dynamic + carveout
+          - 12
+          - 32 or 64
+          - 128
+          - N/A
+          - 2
+          - 256
+          - 32
+          - 16
+          - 32
+          - 512
+          - 32
+          - 11
+          - 0
+        *
+          - AMD Ryzen 9 270
+          - Radeon 780M
+          - RDNA3
+          - gfx1103
+          - Dynamic + carveout
+          - 12
+          - 32 or 64
+          - 128
+          - N/A
+          - 2
+          - 256
+          - 32
+          - 16
+          - 32
+          - 512
+          - 32
+          - 11
+          - 0
+        *
+          - AMD Ryzen AI 9 HX 375
+          - Radeon 890M
+          - RDNA3.5
+          - gfx1150
+          - Dynamic + carveout
+          - 16
+          - 32 or 64
+          - 128
+          - N/A
+          - 2
+          - 256
+          - 32
+          - 16
+          - 32
+          - 512
+          - 32
+          - 11
+          - 5
+        *
+          - AMD Ryzen AI Max+ PRO 395
+          - Radeon 8060S
+          - RDNA3.5
+          - gfx1151
+          - Dynamic + carveout
+          - 40
+          - 32 or 64
+          - 128
+          - 32
+          - 2
+          - 256
+          - 32
+          - 16
+          - 32
+          - 768
+          - 32
+          - 11
+          - 5
+        *
+          - AMD Ryzen Al 7 350
+          - Radeon 860M
+          - RDNA3.5
+          - gfx1152
+          - Dynamic + carveout
+          - 8
+          - 32 or 64
+          - 128
+          - N/A
+          - 1
+          - 256
+          - 32
+          - 16
+          - 32
+          - 512
+          - 32
+          - 11
+          - 5
+
 Glossary
 ========
 

docs/sphinx/_toc.yml.in

@@ -29,27 +29,25 @@ subtrees:
     title: Deep learning frameworks
     subtrees:
     - entries:
-      - file: compatibility/ml-compatibility/pytorch-compatibility
+      - file: compatibility/ml-compatibility/pytorch-compatibility.rst
         title: PyTorch compatibility
-      - file: compatibility/ml-compatibility/tensorflow-compatibility
+      - file: compatibility/ml-compatibility/tensorflow-compatibility.rst
         title: TensorFlow compatibility
-      - file: compatibility/ml-compatibility/jax-compatibility
+      - file: compatibility/ml-compatibility/jax-compatibility.rst
         title: JAX compatibility
-      - file: compatibility/ml-compatibility/verl-compatibility
+      - file: compatibility/ml-compatibility/verl-compatibility.rst
         title: verl compatibility
-      - file: compatibility/ml-compatibility/stanford-megatron-lm-compatibility
+      - file: compatibility/ml-compatibility/stanford-megatron-lm-compatibility.rst
         title: Stanford Megatron-LM compatibility
-      - file: compatibility/ml-compatibility/dgl-compatibility
+      - file: compatibility/ml-compatibility/dgl-compatibility.rst
         title: DGL compatibility
-      - file: compatibility/ml-compatibility/megablocks-compatibility
+      - file: compatibility/ml-compatibility/megablocks-compatibility.rst
         title: Megablocks compatibility
-      - file: compatibility/ml-compatibility/taichi-compatibility
+      - file: compatibility/ml-compatibility/ray-compatibility.rst
-        title: Taichi compatibility
-      - file: compatibility/ml-compatibility/ray-compatibility
         title: Ray compatibility
-      - file: compatibility/ml-compatibility/llama-cpp-compatibility
+      - file: compatibility/ml-compatibility/llama-cpp-compatibility.rst
         title: llama.cpp compatibility
-      - file: compatibility/ml-compatibility/flashinfer-compatibility
+      - file: compatibility/ml-compatibility/flashinfer-compatibility.rst
         title: FlashInfer compatibility
   - file: how-to/build-rocm.rst
     title: Build ROCm from source
@@ -77,8 +75,14 @@ subtrees:
         - entries:
           - file: how-to/rocm-for-ai/training/benchmark-docker/primus-megatron.rst
             title: Train a model with Primus and Megatron-LM
+            entries:
+            - file: how-to/rocm-for-ai/training/benchmark-docker/megatron-lm.rst
+              title: Train a model with Megatron-LM
           - file: how-to/rocm-for-ai/training/benchmark-docker/primus-pytorch.rst
             title: Train a model with Primus and PyTorch
+            entries:
+            - file: how-to/rocm-for-ai/training/benchmark-docker/pytorch-training.rst
+              title: Train a model with PyTorch
           - file: how-to/rocm-for-ai/training/benchmark-docker/jax-maxtext.rst
             title: Train a model with JAX MaxText
           - file: how-to/rocm-for-ai/training/benchmark-docker/mpt-llm-foundry
@@ -117,6 +121,8 @@ subtrees:
             title: SGLang inference performance testing
           - file: how-to/rocm-for-ai/inference/benchmark-docker/sglang-distributed.rst
             title: SGLang distributed inference with Mooncake
+          - file: how-to/rocm-for-ai/inference/xdit-diffusion-inference.rst
+            title: xDiT diffusion inference
           - file: how-to/rocm-for-ai/inference/deploy-your-model.rst
             title: Deploy your model
 
@@ -134,6 +140,8 @@ subtrees:
             title: Profile and debug
           - file: how-to/rocm-for-ai/inference-optimization/workload.rst
             title: Workload optimization
+          - file: how-to/rocm-for-ai/inference-optimization/vllm-optimization.rst
+            title: vLLM V1 performance optimization
 
       - url: https://rocm.docs.amd.com/projects/ai-developer-hub/en/latest/
         title: AI tutorials
@@ -180,7 +188,7 @@ subtrees:
           - file: conceptual/gpu-arch/mi300-mi200-performance-counters.rst
             title: MI300 and MI200 performance counters
           - file: conceptual/gpu-arch/mi350-performance-counters.rst
-            title: MI350 series performance counters
+            title: MI350 Series performance counters
       - file: conceptual/gpu-arch/mi250.md
         title: MI250 microarchitecture
         subtrees:
@@ -214,6 +222,8 @@ subtrees:
     title: ROCm tools, compilers, and runtimes
   - file: reference/gpu-arch-specs.rst
   - file: reference/gpu-atomics-operation.rst
+  - file: reference/env-variables.rst
+    title: Environment variables
   - file: reference/precision-support.rst
     title: Data types and precision support
   - file: reference/graph-safe-support.rst

docs/what-is-rocm.rst

@@ -123,7 +123,8 @@ Performance
 
 .. note::
 
-  `ROCprof Compute Viewer <https://rocm.docs.amd.com/projects/rocprof-compute-viewer/en/amd-mainline/>`_ is a tool for visualizing and analyzing GPU thread trace data collected using :doc:`rocprofv3 <rocprofiler-sdk:index>`. Note that `ROCprof Compute Viewer <https://rocm.docs.amd.com/projects/rocprof-compute-viewer/en/amd-mainline/>`_ is in an early access state. Running production workloads is not recommended.
+  `ROCprof Compute Viewer <https://rocm.docs.amd.com/projects/rocprof-compute-viewer/en/amd-mainline/>`_ is a tool for visualizing and analyzing GPU thread trace data collected using :doc:`rocprofv3 <rocprofiler-sdk:index>`.
+  Note that `ROCprof Compute Viewer <https://rocm.docs.amd.com/projects/rocprof-compute-viewer/en/amd-mainline/>`_ is in an early access state. Running production workloads is not recommended.
 
 Development
 ^^^^^^^^^^^