Commit Graph

4779 Commits

Author SHA1 Message Date
iliya mironov
eefaf56075
Fix unit tests for select layer. (#638)
* Fix unit tests for select layer.
2020-06-08 18:39:40 +03:00
Maxim Vafin
f1811ad060
Implement support for opset3 EmbeddingBag ops (#546)
* [MO] Implement EmbeddingBag_3

* Transform dynamic sub-graph of Wide and Deep into EmbeddingSegmentsSum

- Expressed SparseWeightedSum sub-graph through EmbeddingSegmentsSum
- Removed experimental SparseWeightedSum layer
- Implemented tests for the transformation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix EmbeddingBag shape infer

* Fix EmbeddingSegmentsSum transformation for Wide and Deep

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix EmbeddingSegmentSum replacer after ports swap

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Update package_BOM.txt

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Add unit tests for EmbeddingXXX shape infer

* Fix ATen resolver

* Remove deleted files from BOM

* Add opset version to embedding_bag

* Use base class for EmbeddingBag

* Fix per_sample_weights case

* Fix EmbeddingSegmentsSum transformation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix EmbeddingBag checks

* Fix ATen front transformation and merge conflicts

* Fix BOM

* Work around limitation for I64 input of W&D model

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Cleanup where operation to fix affect of WhereDecomposition transform

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix BOM

* Correct EmbeddingSegmentSum transform for Wide and Deep

Add casting segment ids to i32 and remove ConstToResult sub-graph.

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Update BOM with RemoveConstToResult transform

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Add more comments for RemoveConstToResult transformation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Remove useless logging in EmbeddingSegmentsSum transformation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Small fixes

* Move EmbeddingBag resolving back to front phase

* Improve error messages

* Fix typo in unittests

* Reimplement sparse_reshape middle transform

Avoid deprecated API.

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Clean-up graph after sparse_reshape and ConstToResult transformation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix clean-up for transformations

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix clean-up for transformation #2

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

Co-authored-by: Roman Kazantsev <roman.kazantsev@intel.com>
2020-06-08 18:06:40 +03:00
Konrad Dobros
d155483573
[IE CLDNN] Optimize 1x1 imad convolution kernel (#757) 2020-06-08 16:44:50 +03:00
Andrey Somsikov
626bc4f3d4
Add commit links to memcheck report (#820) 2020-06-08 15:08:58 +03:00
Tomasz Dołbniak
60d4d62536
Disable warnings-as-errors for ONNX target (#749)
* Disable warnings-as-errors for ONNX target

* Disable warnigs-as-errors for windows too

* Change WIN32 -> MSVC
2020-06-08 13:52:45 +03:00
Evgenya Stepyreva
e7f5f53f92
[ MO ] Groupped conv fusion (#797)
Fixed the group convolution fusion pass to properly get the feature dim in NCHW layout case.
2020-06-08 13:00:54 +03:00
Edward Shogulin
a224078c5c
[LPT] [Test] DepthToSpace sporadic fail fix (#815) 2020-06-08 12:55:37 +03:00
Alexander Zhogov
9a968b12db
Azure CI: increase timeout for Mac to 180 min 2020-06-08 12:17:50 +03:00
Vladimir Paramuzov
f0498ad011
[IE CLDNN] Enable ShuffleChannels op (#787) 2020-06-07 22:57:20 +03:00
Edward Shogulin
63ee9f8916
[LPT] [CPU] DepthToSpace transformation (#663)
* [LPT] [TEST] LayerTransformation generalization

* [LPT] DequantizationDetails extending

* [LPT] DepthToSpace transformation implementation
2020-06-07 21:12:52 +03:00
Alexander Zhogov
93b60cacfa
Azure: Add Ninja (#803)
* Azure: Add Ninja

* Fix 'Install Ninja' on Linux

* Fix bin dir path on Windows

* Add -Wno-unused-variable on Mac

* Add -Wno-error=unused-command-line-argument on Mac

* Set CXXFLAGS for Mac

* Improvements

* Fix BIN_DIR on Linux
2020-06-06 15:56:24 +03:00
Vladimir Paramuzov
0022eebd71
[IE CLDNN] Enable DepthToSpace (#780)
Enabled DepthToSpace ngraph transformat
Updated implementation to support 5d and mode parameter
fsv16 direct support
Functional tests for GPU
2020-06-05 20:16:47 +03:00
Daria Mityagina
807f85f93f
it is duplicate of PR: #656, but without test modification (#794) 2020-06-05 19:57:03 +03:00
Chance Luo
7f09b54af8
Disable Hw Avg Pooling for small output tensors if excludePad=true (#772) 2020-06-05 19:47:53 +03:00
Kami-996
cad3ccd8a3
add /wd4819 to disable C4819 warning, which is treated as error in win32 (#767)
Co-authored-by: jasonlee <jasonlee@qiyi.com>
2020-06-05 16:04:59 +03:00
emmanuelattia-philips
a0d1dae91d
Fix: ITT_INCLUDE_DIR was not correctly detected (#748) 2020-06-05 14:46:39 +03:00
Ilya Znamenskiy
4d3ddc1684
[IE CLDNN] GEMM int8 optimization using MMAD macro (#635) 2020-06-05 14:28:21 +03:00
Anton Voronov
70c2058b61
[CPU] supported ShuffleChannels and added tests (#636) 2020-06-05 14:10:55 +03:00
Ilya Churaev
3571d44896
Save the name of output data if we remove previous layer (#760)
* Save the name of output data if we remove previous layer

* Added test
2020-06-05 13:36:35 +03:00
Pavel Rodionov
20d812d959
[GNA] Set default GNA library to GNA2 (#771) 2020-06-05 13:00:58 +03:00
dmitrygo
b485e829d6 [CPU] DepthToSpace review leftovers 2020-06-05 12:47:24 +03:00
Maxim Vafin
f51c533ea8
Add ReduceL2 decomposition (#733)
* Add ReduceL2 decomposition

* Add ReduceL2 transformation tests

* Add const propagation unit test for ReduceL2
2020-06-05 12:34:57 +03:00
Denis Orlov
67e3e06bee
Fix hetero mode in speech sample - set config when loading network (#786) 2020-06-05 11:54:03 +03:00
Ilya Churaev
7a5d447e9f
[SAMPLES] Use defined constant instead of string (#788) 2020-06-05 11:22:24 +03:00
Gladilov, Gleb
f80bd537bf
[IE][VPU][nGraph]: Fixes DTS transformations to properly keep outputs names (#734)
* NonZero, Broadcast

* Concat

* Gather

* [IE][VPU][nGraph]: Fixes DTS transformations to correctly keep outputs names

* [IE][VPU][nGraph]: Fixes dynamic to static shape nonzero tests

Co-authored-by: Roman Vyunov <roman.vyunov@intel.com>
2020-06-05 11:16:52 +03:00
Edward Shogulin
f9ac555857
[LPT] Output layers update fix (#754) 2020-06-05 10:54:38 +03:00
Sergey Shlyapnikov
6e491a89ad
[IE CLDNN] Improve Gather performance and add fusing support (#736) 2020-06-05 10:20:58 +03:00
Egor Churaev
2100521a14
[IE CLDNN] Implement NormalizeL2 int8 kernels (#720) 2020-06-05 10:16:27 +03:00
Ilya Churaev
a705f0c358
Avoid loading of reader if it doesn't exist (#758)
* Avoid loading of reader if it doesn't exist

* Updated error messages
2020-06-04 21:21:13 +03:00
Maxim Vafin
c7d130efbe
Fix Proposal for the case of 2 outputs (#773) 2020-06-04 20:56:46 +03:00
Evgeny Lazarev
c10ff28f12
Added default value for 'aligned' in the ExperimentalDetectronROIFeatureExtractor for backward compatibility (#777)
Fixed backward compatibility issue that old IRs with ExperimentalDetectronROIFeatureExtractor operation cannot be loaded with the new IE
2020-06-04 20:47:52 +03:00
Lukasz Debski
698dfc4bf6
[IE CLDNN] Permute fused ops support (#642) 2020-06-04 17:01:21 +03:00
Alexey Varyzgin
85aa23ec8a
[CPU][BF16] Default Optimisation Capability of BF16 was enabled on CPX (#647) 2020-06-04 16:06:15 +03:00
Maxim Vafin
1001caf04e
Add support for ONNX Pad-11 (#744) 2020-06-04 14:48:31 +03:00
Denis Orlov
0e60aed97a
[GNA] Support 100 inputs, instead of 10 (#741) 2020-06-04 14:33:09 +03:00
Gorokhov Dmitriy
3183c116d9
DepthToSpace, SpaceToDepth layers optimizations (#706)
* [CPU] Updated DepthToSpace and SpaceToDepth layers to be conformant with the specification

The patch also includes n[d]hwc layout support as well as some optimizations

* [CPU][TESTS] Removed old DepthToSpace test since it doesn't corresponds to layer's specification

* [nGraph] Utilize CommonOptimizations pass with custom transformations callback
2020-06-04 14:25:19 +03:00
Evgenya Stepyreva
01e60d057d
[ MO ] InterpolateConcat empty sources fix (#764) 2020-06-04 14:18:33 +03:00
Vladimir Paramuzov
d7fad0109a
[IE CLDNN] Disabled sporadic detection output tests (#740) 2020-06-04 11:14:05 +03:00
Vladimir Paramuzov
28ffbf0857
[IE CLDNN] Remove unused fused deps for FQ (#712)
Remove unused fused FQ kernel arguments to avoid extra setArg() calls which significantly reduces host overhead
2020-06-04 10:30:46 +03:00
Egor Churaev
546377dc8e
[IE CLDNN] Implement EmbeddingBag operations (#623)
Implemented three operations: EmbeddingBagPackedSum,
EmbeddingBagOffsetsSum and EmbeddingSegmentsSum. These operations do
the same work but have a different format of inputs.
2020-06-04 10:25:28 +03:00
Anton Voronov
e53b1b7fbc
[MKLDNN_PLUGIN] Convolution node: skip initializing of primitive descriptors for planar layout if there is already jit primitive (#672) 2020-06-04 08:06:14 +03:00
Ilya Lavrenov
158d32139f
Revert "Enabled thread tests (#717)" (#756)
This reverts commit 99a2423ec0.
2020-06-03 22:32:55 +03:00
wistal
2bb7010193
MO should support LRN k param with caffe model, rather than fixed to 1 (#716)
Co-authored-by: yipengqu <yipeng.qu@intel.com>
2020-06-03 20:33:55 +03:00
Alexey Suhov
1ffada0b23
[Docs] Fixes in readme files: (#750)
- change repo name to openvino
- update driver version
- fix path to samples data
- remove section about Movidius driver installation
- change latest release to 2020.3
- merge fixes in install_dependencies.sh from 2020 branch
2020-06-03 20:14:35 +03:00
Mikołaj Życzyński
023344a317
[IE CLDNN] Added fusing suport to all pooling kernels (#689)
adds fusing support to all available pooling kernels
tests all possible input type/output type configurations
fixes minor bug in max pooling in pooling_gpu_test.cpp
fixed minor bug with yxbf format in pooling_gpu_ref and pooling_gpu_int8_ref kernels
fixes bug with b_fs_yx_fsv32 format in pooling_gpu kernel
resolves bug with max pooling accuracy missmatch in case of non zero pad end layer parameter
resolves average pooling accuracy missmatch in case of non zero pad end layer parameter
2020-06-03 19:44:27 +03:00
Lukasz Debski
e2d1ae7055
[IE CLDNN] Fixed stack overflow in calculate_prior_boxes pass (#747)
The problem behind this error was in program_impl::init_graph() where in calculate_prior_boxes we are trying to calculate output layout of an entire network recursively which causes stack overflow. Calculating output layouts beforehand in processing order fixes this issue.
2020-06-03 19:42:50 +03:00
Ilya-Krylov
cfb5f27899
Add 'aligned' param to ExperimentalDetectronROIFeatureExtractor for CPU plugin and MO 2020-06-03 17:52:40 +03:00
Tomasz Dołbniak
53927034da
Python API for Assign, ReadValue and ExtractImagePatches (#719) 2020-06-03 15:01:43 +02:00
LiweiSong
63a77bb4a1
mkldnn_memory_solver.hpp: include stdint.h to avoid build error (#729)
fix the following compile error:

inference-engine/src/mkldnn_plugin/mkldnn_memory_solver.hpp:60:9: error: 'int64_t' does not name a type
|    60 |         int64_t size;
|       |         ^~~~~~~

include stdint.h to fix this.

Signed-off-by: Liwei Song <liwei.song@windriver.com>
2020-06-03 15:19:29 +03:00
Edward Shogulin
7edebd8d87
[LPT] [TEST] Sporadic test fail fix (workaround) (#742) 2020-06-03 15:05:45 +03:00