github/concrete
1
1
Fork 0
mirror of https://github.com/zama-ai/concrete.git synced 2026-02-09 12:15:09 -05:00
Code Issues Packages Projects Releases Wiki Activity
713 Commits 185 Branches 425 Tags
cc1202de7685aefb6dd68dc0a7e587650bc06e28
Commit Graph

413 Commits

Author SHA1 Message Date
Quentin Bourgerie
9257404f5f fix(python-bindings): Support np.array with dtype upt to 64 bits 2022-08-19 14:15:21 +02:00
Quentin Bourgerie
d647bc735f fix: Now the maximum precision is 16bits 2022-08-16 08:40:14 +02:00
Quentin Bourgerie
5492ed01eb feat: plug the compiler pipeline with the optimizer for 16 bits eint 2022-08-16 08:40:14 +02:00
rudy
f3933aeddc feat(optimizer): connect to new entry point supporting woppbs 2022-08-16 08:40:14 +02:00
Antoniu Pop
93802d128b fix(dfr-compiler): clone memref task arguments using identity maps for serialization. 2022-08-12 16:35:11 +02:00
Quentin Bourgerie
8cd3a3a599 feat(compiler): First draft to support FHE.eint up to 16bits 
For now what it works are only levelled ops with user parameters. (take a look to the tests)

Done:
- Add parameters to the fhe parameters to support CRT-based large integers
- Add command line options and tests options to allows the user to give those new parameters
- Update the dialects and pipeline to handle new fhe parameters for CRT-based large integers
- Update the client parameters and the client library to handle the CRT-based large integers

Todo:
- Plug the optimizer to compute the CRT-based large interger parameters
- Plug the pbs for the CRT-based large integer
 2022-08-12 16:35:11 +02:00
rudy
cc6c2576ec feat(optimizer): create optimizer dag and use it 2022-08-11 10:10:27 +02:00
rudy
bd3d462384 feat(multiprecision): enable real multiple precision computation 2022-08-11 10:10:27 +02:00
Quentin Bourgerie
534e683055 fix: Lowering from ConcreteToBConcrete of from_elements on ND tensor of Concrete.lwe_ciphertext 2022-08-09 10:26:32 +02:00
Antoniu Pop
a2ff61d1d7 cleanup(compiler): remove no longer needed function getAliasedUses. 2022-08-04 17:06:09 +01:00
Antoniu Pop
f9ce4d046b fix(testing-dfr): fix startup and ternimation call parameters to allow sequential test execution on root node when the runtime is instantiated for distributed execution. 2022-08-04 17:06:09 +01:00
Antoniu Pop
f541af8a8f fix(compiler): use ViewLikeOpInterface instead of checking specific op types. 2022-08-04 17:06:09 +01:00
Antoniu Pop
f051642f1f cleanup(compiler): rename getSizeInBytes function to match updated functionality. 2022-08-04 17:06:09 +01:00
Antoniu Pop
46b1dbcd34 cleanup(compiler): fix unused variable warning. 2022-08-04 17:06:09 +01:00
Antoniu Pop
20d19c2161 cleanup(dfr): remove unused key ids and unranked memref labels. 2022-08-04 17:06:09 +01:00
Antoniu Pop
da9dbcef1d feat(compiler): coarsen task granularity by aggregating lightweight operations. 2022-08-04 17:06:09 +01:00
Antoniu Pop
ba60d74da0 feat(dfr): add timing measurements. 2022-08-04 17:06:09 +01:00
Antoniu Pop
1bb3d04059 fix(dfr): broadcast evaluation keys early to avoid locking in HPX helper threads. 2022-08-04 17:06:09 +01:00
Antoniu Pop
fbca52f4a0 feat(dfr): add memory management for futures and associated data. 2022-08-04 17:06:09 +01:00
Antoniu Pop
b405a2daf2 fix(dfr): fix thread binding when running both HPX and OpenMP runtimes. 2022-08-04 17:06:09 +01:00
Antoniu Pop
eeb3ba8735 fix(compiler): add calls to enable either or both parts of the runtime based on compile options used. 2022-08-04 17:06:09 +01:00
Antoniu Pop
51d3959742 fix(dfr): remove runtime resume/suspend on entry/exit of function as currently unstable. 2022-08-04 17:06:09 +01:00
Antoniu Pop
49111dd295 fix(dfr): simplify runtime parametrization and fix implicit thread allocation policy. 2022-08-04 17:06:09 +01:00
Antoniu Pop
ae55e88435 feat(dfr): add default configuration file for HPX runtime. 2022-08-04 17:06:09 +01:00
Antoniu Pop
7410502e60 fix(dfr): simplify synchronization in work function registry. 2022-08-04 17:06:09 +01:00
Antoniu Pop
e36c7d3b55 feat(dfr): add environment variable option for using HPX configuration file path at initialization. 2022-08-04 17:06:09 +01:00
Antoniu Pop
615109d432 fix(dfr): simplify initialization and add namespaces where possible. 2022-08-04 17:06:09 +01:00
Antoniu Pop
460fbabbe0 feat(runtime): deactivate main wrapping by default and add explicit initialization/termination. 2022-08-04 17:06:09 +01:00
Antoniu Pop
bca85ea2b6 feat(compiler): delay the point where futures are waited on, when they are being synchronized back to their parent context, to the point of use. 2022-08-04 17:06:09 +01:00
Antoniu Pop
954b2098c6 feat(runtime): enable distributed execution. 2022-08-04 17:06:09 +01:00
Antoniu Pop
2cc8c69ff3 fix(compiler): make the insertForwardDeclaration utility function accept any OpBuilder for more generality. 2022-08-04 17:06:09 +01:00
Antoniu Pop
1df468a084 feat(runtime): increase the limit of task input parameters to 16. 2022-08-04 17:06:09 +01:00
youben11
ce9393070c fix: avoid tensor slices in conv operation 
when lowering the linalg conv operation into scf, then bufferized, there
is a cast operation on an allocated buffer to the type of a
memref with different strides (casued by the slice). This incompatible
cast makes the lowering fails.
 2022-08-04 11:19:18 +01:00
youben11
e52ccfc1a9 feat: support grouped conv2d 2022-08-04 11:19:18 +01:00
youben11
63d84a3e4a refactor: remove code related to our custom conv2D named op 
it was introduced as a workaround while linalng couldn't support other
types than int/float
 2022-08-04 11:19:18 +01:00
youben11
f1f1db923d refactor: lower our conv2d to custom linalg named op 
we can now generate linalg named op with custom operation for add/mul to
handle our types
 2022-08-04 11:19:18 +01:00
Andi Drebes
85ebc0cb7a Rebase onto llvm-project 3f81841474fe with patch for arbitrary types in linalg named ops 
Rebase to llvm-project at 3f81841474fe with a pending upstream patch
for arbitrary element types in linalg named operations.

Co-authored-by: Ayoub Benaissa <ayoub.benaissa@zama.ai>
 2022-07-27 22:45:38 +02:00
Andi Drebes
fb9066c3e8 fix(compiler): Fix calculation of indexes for result memrefs with non-zero offsets 
The function `tensorDataFromMemRef` reading elements from a result
memref and copying them to an instance of `TensorData` fails to handle
memrefs with non-zero offsets as it adds the offset twice when
calculating indexes. This may result in out-of-bounds memory accesses
and incorrect results.

This patch fixes the calculation of indexes by removing the second
addition of the offset.
 2022-07-27 22:39:54 +02:00
Quentin Bourgerie
511bcd99e7 Revert "feat(optimizer): create optimizer dag and use it" 
This reverts commit 0b99f6d278.
 2022-07-27 18:35:47 +02:00
rudy
0b99f6d278 feat(optimizer): create optimizer dag and use it 2022-07-25 21:06:14 +02:00
Umut
e24dbec249 feat: create FHELinalg.from_element operation 
This commit is introduced because python bindings for `tensor.from_elements` are not generated automatically. Previously, we overcame this with string manipulation, but with the latest version of the compiler, it became a problem. This commit should be reverted eventually. See https://discourse.llvm.org/t/cannot-create-tensor-from-elements-operation-from-python-bindings/4768 for the discussion in LLVM forums.
 2022-07-08 14:43:46 +03:00
youben11
f4166a4973 docs: use consistent style for comment blocks 
prefix comment blocks with ///
 2022-07-07 16:11:19 +01:00
Antoniu Pop
261e59d2b4 feat(dfr): add more operations to be converted to tasks. 2022-06-28 17:11:12 +01:00
Antoniu Pop
26084a68aa fix(compiler): delay all memref deallocation calls introduced by the bufferizer and that are made into futures until after the synchronization point. 2022-06-28 17:11:12 +01:00
Antoniu Pop
fb219f0fbf feat(runtime): increase the limit of task input parameters to 16. 2022-06-28 17:11:12 +01:00
Umut
b3a2671dc7 feat: implement all kinds of subtractions 2022-06-28 11:13:03 +03:00
Umut
8f8a57d220 feat: implement MANP calculation for tensor.insert op 2022-06-28 11:13:03 +03:00
Mayeul@Zama
bcc5c750a1 fix(client): fix assumption that glweDimension = 1 2022-06-28 09:24:39 +02:00
Quentin Bourgerie
2de76e9c4e fix(compiler): Fix tfhe global parametrization to handle k>1 
Co-authored-by: Mayeul@Zama <mayeul.debellabre@zama.ai>
 2022-06-24 15:35:51 +02:00
Quentin Bourgerie
ef9d11c16f cleanup(compiler): Remove dead code and usless pass 2022-06-24 11:00:08 +02:00