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
117 Commits 185 Branches 425 Tags
ea3c940f4a79d5df823859aa04dab2dabcfc1ffd
Commit Graph

63 Commits

Author SHA1 Message Date
youben11
ea3c940f4a feat: lower LowLFHE types (apply_lut) to llvm.ptr 2021-09-09 20:35:28 +02:00
youben11
6e2ac3af4e feat: parameterize KS operation 2021-09-09 20:35:28 +02:00
youben11
14f171bef9 feat: parameterize bst operation 
This is needed to have the right parameters when calling concrete
 2021-09-09 20:35:28 +02:00
youben11
2b0cfce1f6 feat(compiler): LowToCAPI of apply_lut 2021-09-09 20:35:28 +02:00
Quentin Bourgerie
d03adbd7dc enhance(compiler): Add options for verbose passes 2021-09-08 11:44:51 +02:00
Quentin Bourgerie
1077c9167c enhance(compiler): Full lowering of hlfhe.zero, make dot_eint_int fully lowerable 2021-09-08 11:44:51 +02:00
Quentin Bourgerie
de7129fe8e feat(compiler/llvm-pipeline): Bufferize linalg and convert to loops 2021-09-08 11:44:51 +02:00
Quentin Bourgerie
a654fb2d0e enhance(compiler/midlfhe): Handle linalg.generic in parametrization pass 2021-09-08 11:44:51 +02:00
Quentin Bourgerie
697d4033e1 fix(compiler): typo 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
19f1a22b6a fix(compiler/midlfhe): Manage tensor in midlfhe parametrization 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
af0789f128 enhance(compiler/runtime): Add runtime tools to handle tensor inputs and outputs 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
dba76a1e1b enhance(compiler): Add tensor ops type rewriting on high level pipepline 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
6ac882fc65 fix(compiler/capi): Remove the memref for error and no more bare pointer call convention 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
bc975d904e feat(compiler): introduce bufferization passes in lowering pipeline to llvm 2021-09-06 10:26:40 +02:00
Quentin Bourgerie
19704b4b13 fix(compiler/lowlfhe): Use the lwe ciphertext parameters to allocate instead of global fhe context 2021-08-24 16:29:04 +02:00
Quentin Bourgerie
4ae00963ac cleanup: Remove unused import 2021-08-24 16:21:31 +02:00
Quentin Bourgerie
67f0fc0f45 enhance(compiler): Introduce MidLFHE dag parametrization 2021-08-24 16:21:31 +02:00
Quentin Bourgerie
70fb5fcd8e fix(compiler/midlfhe): Change constraint on operators with integers (just too large integers are forbidden) 2021-08-24 16:21:31 +02:00
Quentin Bourgerie
8057ee7553 refactor(compiler): Prepare the MidLFHE parameters injection 2021-08-24 16:21:31 +02:00
youben11
f948db1228 feat(python): CompilerEngine to compile and run 2021-08-18 17:38:46 +02:00
Quentin Bourgerie
5613c69602 cleanup(compiler): Remove debug logs... 2021-08-18 17:06:28 +02:00
Quentin Bourgerie
fa62e1f0e5 refactor(compiler): Move memref HLFHE and MidLFHE operators to tensor 2021-08-18 11:15:30 +02:00
Quentin Bourgerie
7372cd3d0a fix(compiler/hlfhe): More verification on dot_eint_int 2021-08-17 16:53:32 +02:00
Andi Drebes
8b9c9f2da1 refactor(compiler): HLFHE.dot_eint_int: Switch from reference to value semantics 
This changes the semantics of `HLFHE.dot_eint_int` from memref-based
reference semantics to tensor-based value semantics. The former:

  "HLFHE.dot_eint_int"(%arg0, %arg1, %arg2) :
     (memref<Nx!HLFHE.eint<0>>, memref<Nxi32>, memref<!HLFHE.eint<0>>) -> ()

becomes:

  "HLFHE.dot_eint_int"(%arg0, %arg1) :
     (tensor<Nx!HLFHE.eint<0>>, tensor<Nxi32>) -> !HLFHE.eint<0>

As a side effect, data-flow analyses become much easier. With the
previous memref type of the plaintext argument it is difficult to
check whether the plaintext values are statically defined constants or
originate from a memory region changed at execution time (e.g., for
analyses evaluating the impact on noise). Changing the plaintext type
from `memref` to `vector` makes such analyses significantly easier.
 2021-08-17 16:53:32 +02:00
Quentin Bourgerie
8796754513 fix(compiler/lowlfhe): Fix wrong shift value for EncodeIntOp lowering 
That's work before because we enfoce 6 as global constraint while our runJit tests are on 7 bits
 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
ed7dd36e70 fix(compiler): Remove the type conversion workaround, actually the order of addConversion calls matters (lifo) 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
42f12b22da fix(compiler): move the midlfhe to lowlfhe pass to the right place 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
f45abce604 feat(compiler): More lowlfhe operators conversion to mlir std dialects 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
03297fd50d fix(compiler/lowlfhe): for the v0 give the lweSize of ciphertext as a global parameter of the lowering pass to concrete api call (#62) 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
b22f585380 feat(compiler): First draft of lowering from LowLFHE to std with fct call (#62) 2021-08-16 18:54:56 +02:00
Quentin Bourgerie
d0877536ed feat(compiler): First draft of client parameters generation, runtime support for encrypting and decrypting circuit gates, integration of fhe parameters for the v0 (#65, #66, #56) 2021-08-13 14:05:29 +02:00
youben11
4d3c7a68de feat(compiler): MidToLowLFHE lowering 2021-08-12 13:48:51 +01:00
youben11
3d8d5c438a refactor(compiler): generalize LinalgGenericPattern 
also fixes an issue regarding populateWithGenerated, which can be
duplicated across different pattern files. So I redefined a different
function that is more unique to the pass that should be ran, and hide
the populateWithGenerated from the global namespace
 2021-08-12 13:48:51 +01:00
Ayoub Benaissa
7a2511b3d4 feat(compiler): LowLFHE dialect (#77) 
* feat(compiler): low level fhe dialect

* feat(compiler): using generated printer/parser in LowLFHE

* feat(compiler): new types and ops for LowLFHE

* tests(compiler): LowLFHE types and ops

* feat(compiler): fill ops

* cleanup

* summary + description

* tests(compiler): use new CLI args

* formatting
 2021-08-05 14:32:46 +01:00
Quentin Bourgerie
b4e57984b1 feat(compiler): Add passes to lower mlir to mlir llvm ir and run jit and emit llvm code (#63) 2021-08-05 13:43:08 +01:00
Quentin Bourgerie
4e6579e019 feat/refactor(compiler): Add --passes options to activate only a subset of passes (#57) 2021-08-04 14:03:05 +02:00
Quentin Bourgerie
1605551f1a feat(compiler): HLFHEToMidLFHE pass (#57) 2021-08-04 14:03:05 +02:00
Quentin Bourgerie
ed40b1ef91 feat(compiler): Add HLFHE.sub_int_eint (#54) 2021-08-03 09:23:11 +02:00
Quentin Bourgerie
143a2384fd enhance(compiler): Add verifier of HLFHE operators 2021-08-03 09:23:11 +02:00
Quentin Bourgerie
cb635f8a55 feat(compiler): HLFHE.apply_lookup_table (#54) 2021-08-03 09:23:11 +02:00
Quentin Bourgerie
ea77ee696a enhance(compiler): Add verify for HLFHE.eint, precision ]0;7] (#54) 2021-08-03 09:23:11 +02:00
Quentin Bourgerie
3c326c09d6 clean(compiler): Remove unecessary operators for compiler v0 (#54) 2021-08-03 09:23:11 +02:00
Quentin Bourgerie
be62755401 fix(compiler): Fixing MidLFHE Dialect for the compiler v0 2021-07-29 09:53:45 +02:00
Ayoub Benaissa
ab53ef71c6 feat(compiler): Python bindings (#53) 
- feat(compiler): python bindings
- build: update docker image for python bindings
- pin pybind11 to 2.6.2, 2.7 is not having correct include_dirs set (still
a question why?)
- using generated parser/printer
 2021-07-28 15:58:51 +01:00
Andi Drebes
9d1cdc6a0c feat(compiler): Add lowering pass from HLFHE.dot_eint_int to linalg.generic 
This pass transforms any instance of `HLFHE.dot_eint_int` to an
instance of `linalg.generic` with an appropriate region using
`HLFHE.mul_eint_int` and `HLFHE.add_eint` operations and an
appropriate specification for the iteration dimensions.

Example:

  "HLFHE.dot_eint_int"(%arg0, %arg1, %arg2) :
    (memref<?x!HLFHE.eint<0>>, memref<?xi32>, memref<!HLFHE.eint<0>>) -> ()

becomes:

  linalg.generic {
    indexing_maps = [affine_map<(d0) -> (d0)>,
                     affine_map<(d0) -> (d0)>,
		     affine_map<(d0) -> ()>],
    iterator_types = ["reduction"]
  } ins(%arg0, %arg1 : memref<?x!HLFHE.eint<0>>, memref<?xi32>) outs(%arg2 : memref<!HLFHE.eint<0>>) {
    ^bb0(%arg3: !HLFHE.eint<0>, %arg4: i32, %arg5: !HLFHE.eint<0>):  // no predecessors
      %0 = "HLFHE.mul_eint_int"(%arg3, %arg4) : (!HLFHE.eint<0>, i32) -> !HLFHE.eint<0>
      %1 = "HLFHE.add_eint"(%0, %arg5) : (!HLFHE.eint<0>, !HLFHE.eint<0>) -> !HLFHE.eint<0>
      linalg.yield %1 : !HLFHE.eint<0>
  }
 2021-07-16 18:03:49 +02:00
Andi Drebes
b433627821 feat(compiler): Add Dot operation to HLFHE dialect 
Add a simple dot product between a vector of encrypted integers and a
vector of plaintext integers to the HLFHE dialect.

The operation takes two input operands and one output operand, all
modeled as memrefs. Example:

  "HLFHE.dot_eint_int"(%lhs, %rhs, %out) :
    (memref<?x!HLFHE.eint<0>>, memref<?xi32>, memref<!HLFHE.eint<0>>) -> ()
 2021-07-16 18:03:49 +02:00
Andi Drebes
cef0e569d4 fix(compiler): Generate and include autogenerated dialect definitions (#52) 
* fix(compiler): Generate and include autogenerated dialect definitions

Since commit 485cc55edfb875628e19bb6d9de4706af2865d3e in llvm-project,
dialects are now required to include a file with autogenerated
definitions. This commit generates the definition files and includes
them in appropriate dialect source files for MidLFHE and HLFHE.

* CI: update docker image ref

Co-authored-by: Andi Drebes <andi.drebes@zama.ai>
Co-authored-by: Ayoub Benaissa <ayoub.benaissa@zama.ai>
 2021-07-08 15:54:10 +01:00
Quentin Bourgerie
3aa5457b84 fix(compiler): Fixing some typos 2021-06-23 10:10:31 +02:00
Quentin Bourgerie
a760ade545 enhance(compiler): Better error message for glwe error 2021-06-23 10:10:31 +02:00
Quentin Bourgerie
c783d85a84 doc(compiler): comments on MidLFHE verifiers 2021-06-23 10:10:31 +02:00