- 29 Aug, 2019 1 commit
-
-
Nagy Mostafa authored
* WIP * Fix incorrect CK output adjustment * Bug fix and enroce sanity check * Change cycle search depth, and fix sanity check * cpu_fusion.validate_fuse_gru_inputs passes. * Fix as_single_output to be able to always create a GOE * minor fix. style-apply * Clean up debug msgs * Switch to backward cycle check * Enable failing test * PR fixes * Address feedback: Add fwd cycle checks. Make cycle checking depth configurable
-
- 28 Aug, 2019 1 commit
-
-
Diego Caballero authored
* [MLIR] Add config files for LIT LIT is the LLVM testing tool that is used in LLVM and MLIR. It's necessary to enable all the MLIR testing in nGraph. This will provide a complementary finer grain level of testing than the one currently done with gtest. LIT will allow testing each MLIR-based nGraph transformation without having to run the whole nGraph pipeline. For example, we will be able to write an nGraph dialect input, invoke an single MLIR transformation on it and check that the output is the expected. This PR adds the basic configuration files to enable LIT tool in nGraph for MLIR Compiler. `affine_lowering/lit_test.mlir` is a dummy test to exercise the new testing feature. Lit testing will be enabled in CMAKE in a subsequent PR. * Address feedback
-
- 26 Aug, 2019 1 commit
-
-
Diego Caballero authored
* [MLIR] Bump MLIR repo 8/20/2019 MLIR_ commit 0cdb20a6add19bc96c20dad28589a1e54e4d8469 Author: Lei Zhang <antiagainst@google.com> Date: Tue Aug 20 13:33:41 2019 -0700 Add spv.specConstant and spv._reference_of LLVM: commit 3b9a27b6908040881dad394022f8c472c15c0784 Author: Simon Pilgrim <llvm-dev@redking.me.uk> Date: Tue Aug 20 17:54:37 2019 +0000 Fix typo in comment. NFCI. * [MLIR] Disable three more tests with MLIR enabled. This PR disables validate_fuse_gru_inputs, reshape_layout_optimizations4 and reshape_layout_optimizations5: 1. trivial_in_place_relu_fail: It checks tensors pool offset. There is no memory pool in MLIR atm. 2. validate_fuse_gru_inputs: It creates an infinite cycle in MLIR subgraph extraction pass (under investigation). 3. reshape_layout_optimizations4/5: They fail due to CompiledKernel being not expected by CPULayout pass. * Disable cpu_quant_fusion.qconcat
-
- 23 Aug, 2019 1 commit
-
-
Nishant Patel authored
* Segregate builders op wise * Style * Update ngraph.hpp
-
- 21 Aug, 2019 2 commits
-
-
Adam Procter authored
* New clang-format rules for src/ngraph/op/util * Restore root .clang-format * Remove redundant .clang-formats * Revert "Remove redundant .clang-formats" This reverts commit e158df1e507570b55348ce326fd7e8b4ae819f36. * Remove redundant .clang-formats
-
Nishant Patel authored
-
- 20 Aug, 2019 5 commits
-
-
Nishant Patel authored
* First pass, remove QDot as experimental op, add it as core op, test case failing * Change QuantizedMatmul API * Tests for QuantizedDot passing, cleanup required * Move test cases to backend_test.in.cpp * Change reference dot API to take scale and zero point * Onnx matmul integer tests passing * Add validate_and_infer_types() * Add more checks for QDot * Add type prop tests * Add interpreter support * Add serializer support * Disable tests in other backends * Fix test and add codegen support * Make axes optional argument * Change fusion pattern * Copyright * merge changes * more merge changes * Fix more merge conflicts * change builders to reflect new changes * cleanup * Add reduction_axes to the API * Fix onnx test cases * Remove debug comments * Fix CI * disable plaidml tests * Remove comment
-
Adam Procter authored
-
Nishant Patel authored
* Remove QuantizedMaxPool * Remove QuantizedAvgPool * Disable tests for plaidML * Fix CI * Add correct casting for int8 type in reference kernel * Disable tests for GPU
-
Robert Kimball authored
* Normalize BatchNorm test names * unit test renamed so manifest must be updated
-
Adam Procter authored
* Dynamic version of builder::flatten * Remove as_single_output_node (no longer needed) * Temp fix for dyn elimination
-
- 19 Aug, 2019 3 commits
-
-
Mateusz Bencer authored
* Changed name from Normalize to Normalize2 * Changed name from Normalize to Normalize2 in tests * Changed name of normalize source files * Removed across spatial and channel shared params * Removed scale and introduced input axes * Support for axes input was introduced * Added possibility to choose method of bias using * Clang style applied * Code review remarks introduced * Code review remarks introduced * Added python script to generate normalize_l2 test data
-
Nagy Mostafa authored
-
Mateusz Bencer authored
* Changed required image_shape type from u64 to float * Proposal op was adjusted * Clang styles applied * Added missing new line * Fixed problem with narrowing conversion * Code review remarks * Removed checking types of Proposal input params * Removed unused input element type variables
-
- 16 Aug, 2019 1 commit
-
-
Nagy Mostafa authored
-
- 15 Aug, 2019 7 commits
-
-
Scott Cyphers authored
-
Pruthvi authored
* - Add graph pass method for onnx lstmcell rewrite with lstm cpu op - insert reshapes to keep the weights in ldigo format - test case for onnx LstmCell to CPU Lstm * fix typo * - check LSTMCell for the fused op decomposistion in the backend * - fix bug in onnx_lstm graph pass - passes unit test * style-fix * - fix compilation error - use IFCO gate ordering for bias * - Skip LSTMCell to LSTM CPU fusion for peephole * - add comment && remove duplicate function * -use dynamic_pointer_cast to check for constant * - onnx bias will be of shape (2 * gates_count * hidden_size) bias of Wb and Rb are concatenated, we will split the bias, add and rearrange in order IFCO * - Use most derived LSTM ctor for pattern matching * - Style Fix * style fix * Address PR comments * - add support for graph pass (MKLDNN version > 1) for mapping LSTMCell -> LSTM CPU op * fix unit test failure for MKLDNN V1.0
-
Tomasz Socha authored
* [FUSED] Change alpha attribute in Elu from dynamic to static * Fix python API * Fix failing python test * Remove unnecesary test * Fix merge artifact * Review fix I * Review Fix PyAPI
-
Scott Cyphers authored
-
Diego Caballero authored
CPU fusion is disabled in MLIR since fused ops are not supported in nGraph dialect. CPU debug tracer test doesn't expect CompiledKernel ops generated for MLIR.
-
Robert Kimball authored
* Add method to check if an Executable can create tensors * Add a Backend check to see if Executable can create tensors * Update per review comment * style
-
Robert Kimball authored
* update fused_clamp * convert all f64 to f32
-
- 14 Aug, 2019 1 commit
-
-
Mateusz Bencer authored
* Axes input was added to LRN * Unit tests for axes shape check were added * LRN node deserialization was updated * Fixed EOF and clang style applied * Changed Constant to Parameter type in unit tests * Fixed LRN assert description * Fixed copy_with_new_args * Clang style applied
-
- 13 Aug, 2019 1 commit
-
-
Dina Suehiro Jones authored
* Update cpu backend to support i32 with topk * Add topk i32 unit test * Minor fix in unit-test * disable test for plaidml
-
- 09 Aug, 2019 3 commits
-
-
Nishant Patel authored
* Add negative op * Add test case * Address feedback * Merge master * Consolidate to one routine for unary ops * Change from Negative to Neg
-
Adam Procter authored
* Fix sum reference to handle corner cases with +-inf * Review comments, and try to make Windows happy * Update GPU unit_test.manifest * More template grindery, to make macOS happy
-
Robert Kimball authored
* wip * parse working with one test * Add version information query and unit tests for parsing version strings * add more comments * doc strings * style
-
- 08 Aug, 2019 1 commit
-
-
Adam Procter authored
-
- 07 Aug, 2019 5 commits
-
-
Denise Kutnick authored
* adding group convolution implementation for plaidml * update CMakeLists.txt to include new group convolution files * adding new ngraph unit tests for grouped convolutions * adding group conv unit test back in for plaidml backend * specify group convolution support in plaidml * add denise as codeowner for plaidml runtime dir * remove commented-out lines of unit test manifest * style changes * resolve all discussions in PR * skip data dilation unit test on cpu backend
-
Denise Kutnick authored
* adding group convolution implementation for plaidml * update CMakeLists.txt to include new group convolution files * adding new ngraph unit tests for grouped convolutions * adding group conv unit test back in for plaidml backend * specify group convolution support in plaidml * add denise as codeowner for plaidml runtime dir * remove commented-out lines of unit test manifest * style changes * resolve all discussions in PR * skip data dilation unit test on cpu backend
-
Mateusz Bencer authored
-
Mateusz Bencer authored
-
Robert Kimball authored
-
- 06 Aug, 2019 1 commit
-
-
Robert Kimball authored
* Unit tests to individual files * rename obsolete file
-
- 05 Aug, 2019 1 commit
-
-
Adam Procter authored
* WIP * CHANGE_DYNAMIC_STATE * Implement full type prop for DynBroadcast when inputs const/static; clean up pass properties * Add a unit test for the late-constness thing * Update reference, interp, constant folding to handle autobroadcast * Document the autobroadcast helper, and fix a corner case (zero's a number too!) * Tests, and insert a check within the CPU folders for autobroadcast (it's still unsupported) * Remove unnecessary include * EXPECT_EQ -> ASSERT_EQ, for consistency * Replace assert for CPU folders with a fallback
-
- 02 Aug, 2019 1 commit
-
-
Adam Procter authored
* WIP * CHANGE_DYNAMIC_STATE * Implement full type prop for DynBroadcast when inputs const/static; clean up pass properties * Add a unit test for the late-constness thing * Fix merge * style
-
- 01 Aug, 2019 4 commits
-
-
Adam Procter authored
-
Adam Procter authored
-
Robert Kimball authored
-
Adam Procter authored
-