Liubov Batanina [Mon, 29 Jun 2020 20:07:08 +0000 (23:07 +0300)]
Fixed Softmax reference (#1148)
Alexander Chaiko [Mon, 29 Jun 2020 17:27:41 +0000 (19:27 +0200)]
[IE CLDNN] Fix Android build error: braces around scalar initializer (#1151)
Ilya Lavrenov [Mon, 29 Jun 2020 13:21:48 +0000 (16:21 +0300)]
Cnnnetwork add layer (#1124)
* Removed addLayer from public interface
* Convert to CNNNetworkImpl
Evgenya Stepyreva [Mon, 29 Jun 2020 11:56:11 +0000 (14:56 +0300)]
[ MO ] Keep data type of compressed value (#1143)
JIRA: 34085
Alexander Chaiko [Mon, 29 Jun 2020 11:09:33 +0000 (13:09 +0200)]
[IE CLDNN] int8 batches optimization (#632)
Vladimir Gavrilov [Mon, 29 Jun 2020 09:49:29 +0000 (12:49 +0300)]
Fixed deleting Transpose layers after and before Interpolate layers. (#1071)
* Fixed deleting Transpose layers after and before Interpolate layers.
* Added run_after() for the transformation InterpolateTranspose.
* Some checks were moved from the replacement function to the pattern.
* Added a check of the attribute 'axes' into the pattern.
Ilya Churaev [Mon, 29 Jun 2020 08:14:48 +0000 (11:14 +0300)]
Removed backprop operations (#1091)
* Removed backprop operations
* Fixed build
* Removed AvgPool
* Removed tests for v0 AvgPool
* Fixed code style
* Fixed export
Egor Churaev [Mon, 29 Jun 2020 07:36:30 +0000 (00:36 -0700)]
[IE CLDNN] Implement ExtractImagePatches operation (#1127)
The ExtractImagePatches operation collects patches from the input
tensor, as if applying a convolution. All extracted patches are stacked
in the depth dimension of the output.
JIRA: 30055
Adam Osewski [Fri, 26 Jun 2020 14:35:00 +0000 (16:35 +0200)]
Dynamic attribute getters and setters. (#964)
Anton Chetverikov [Fri, 26 Jun 2020 10:41:49 +0000 (13:41 +0300)]
Fix const node non-deterministic names (part 1) (#996)
* Update node names
Adam Osewski [Fri, 26 Jun 2020 10:00:28 +0000 (12:00 +0200)]
Fix use of new TestCase. (#1130)
Michał Karzyński [Fri, 26 Jun 2020 08:16:51 +0000 (10:16 +0200)]
Fixes in TensorIterator builder code (#1104)
Adam Osewski [Fri, 26 Jun 2020 07:38:15 +0000 (09:38 +0200)]
[ONNX] Support for dynamic shapes for InstanceNormalization (#1076)
Mateusz Bencer [Fri, 26 Jun 2020 07:18:24 +0000 (09:18 +0200)]
Enabled importing ONNX Yolo v3, added Loop op (#957)
Ilya Lavrenov [Thu, 25 Jun 2020 17:00:39 +0000 (20:00 +0300)]
Removed getLayerByName from public API (#1110)
* Fixed tests
* Removed getLayerByName from public API
Andrey Zaytsev [Thu, 25 Jun 2020 15:46:21 +0000 (18:46 +0300)]
Link fixes for opset docs (#1072)
* Fixed links
* Update opset.md
Ewa Tusień [Thu, 25 Jun 2020 13:24:00 +0000 (15:24 +0200)]
Switch MeanVarianceNormalization op to opset3 in ONNX importer (#865)
* Switch MVN op to opset3.
Anna Alberska [Thu, 25 Jun 2020 09:43:47 +0000 (11:43 +0200)]
[GNA] fix custom scale factors when importing a model (#1096)
Andrew Bakalin [Thu, 25 Jun 2020 07:38:25 +0000 (10:38 +0300)]
[VPU] WA for statis shape allocation (#1107)
Jedrzej Hajduczenia [Thu, 25 Jun 2020 07:24:00 +0000 (09:24 +0200)]
[IE CLDNN] Set strided slice out_format to bfyx when in_format=bfzyx and shrink_axis_mask is set (#1111)
Maxim Shevtsov [Thu, 25 Jun 2020 06:25:56 +0000 (09:25 +0300)]
LayerNorm(PyTorch/HuggingFace pattern)->MVN+Mul+Add (#1003)
* LayerNorm(PyTorch/HuggingFace pattern)->MVN+Mul+Add. Improves perf on BERT by 5%
* deducing the across_channels from axes passed to the MVN op.
axes are normalized. if no axes is specified, falling back to the (previously) default across_channel value
Co-authored-by: myshevts <maim.y.shevtsov@intel.com>
Evgenya Stepyreva [Wed, 24 Jun 2020 20:09:48 +0000 (23:09 +0300)]
[ v4::NMS ] Fixed v4 NMS cloning (#1113)
Ilya Lavrenov [Wed, 24 Jun 2020 15:40:06 +0000 (18:40 +0300)]
Removed suppression macro usage (#1108)
Roman Lyamin [Wed, 24 Jun 2020 15:30:24 +0000 (18:30 +0300)]
[IE CLDNN] Added space_to_batch operation (#984)
Andrey Dmitriev [Wed, 24 Jun 2020 14:38:34 +0000 (17:38 +0300)]
[GNA] Added fix multiple output with one go to memory and test (#669)
[GNA] Added fix multiple output with one go to memory and test
[GNA] Added fix multiple output with one go to memory and test
[GNA] Added fix multiple output with one go to memory and test
Added multi output
Update gna_pass_manager.cpp
test
[GNA] Added fix multiple output with one go to memory and test
[GNA] Added fix multiple output with one go to memory and test
[GNA] Added fix multiple output with one go to memory and test
Added multi output
Update gna_pass_manager.cpp
test
tests
[GNA] Added fix multiple output with one go to memory and test
[GNA] Added fix multiple output with one go to memory and test
Added multi output
Update gna_pass_manager.cpp
test
tests
Added pass
Test
test
tests_2
return old
Tomasz Dołbniak [Wed, 24 Jun 2020 14:02:39 +0000 (16:02 +0200)]
TestCase & TestEngine(s) for nGraph UTs (#934)
Irina Efode [Wed, 24 Jun 2020 13:10:39 +0000 (16:10 +0300)]
[IE TESTS] CoreThreadingTestsWithIterations.smoke_LoadNetworkAccuracy disabled & small refactoring of CoreThreadingTests (#1103)
Ilya Lavrenov [Wed, 24 Jun 2020 12:12:14 +0000 (15:12 +0300)]
Exec extensions (#963)
* Fixes
* Removed some tests for extensions
* Added const
* Removed unknown pragma
Ilya Churaev [Wed, 24 Jun 2020 09:44:19 +0000 (12:44 +0300)]
Removed reference implementations for some data types (#1086)
Irina Efode [Wed, 24 Jun 2020 09:19:11 +0000 (12:19 +0300)]
[IE TESTS] Remove extra files (#1087)
CI passed
Roman Kazantsev [Wed, 24 Jun 2020 09:14:18 +0000 (14:14 +0500)]
[IE] Preserve output data name after merging and update output data map (#1101)
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
Ilya Lavrenov [Wed, 24 Jun 2020 08:55:17 +0000 (11:55 +0300)]
Moved deprecated network iterator to legacy (#913)
* Removed deprecated iterator API
* Applied comments
Ilya Lavrenov [Wed, 24 Jun 2020 02:53:28 +0000 (05:53 +0300)]
Removed deprecated ICNNNetwork::getData (#1093)
Ilya Lavrenov [Tue, 23 Jun 2020 19:34:26 +0000 (22:34 +0300)]
Removed ICNNNetReader interface (#1042)
* Removed ICNNNetReader interface
* Fixed stress tests
* Fixed comments in VPU plugin
* Removed duplicated stress tests
* Fixed watchdog tests
Anton Zaytsev [Tue, 23 Jun 2020 18:43:13 +0000 (21:43 +0300)]
[IE TESTS] move BehaviorTestPlugin to the new IE tests infra & small refactoring for Behavior tests (#784)
* [ci-skip][IE TESTS] move beh_test_plugin
* [ci-skip][IE TESTS] move BehaviorHolderTest
* [ci-skip][IE TESTS] fix GNA layout test
* [ci-skip][IE TESTS] fix cmake
* [ci-skip][IE TESTS] fix lib in IEBehaviorTest
* [ci-skip][IE TESTS] separate layout and cpp_wrapers test and fix namespace
* [ci-skip][IE TESTS] fix holders test
* [ci-skip][IE TESTS] fix namespace
* [ci-skip][IE TESTS] fix codestyle
* [ci-skip][IE TESTS] fix test_plugin
* [ci-skip][IE TESTS] fix test_plugin v2
* [ci-skip][IE TESTS] disabled gpu instance for test_plugin
* [ci-skip][IE TESTS] fix
Irina Efode [Tue, 23 Jun 2020 17:39:06 +0000 (20:39 +0300)]
[IE TESTS] Range (#1094)
Michał Karzyński [Tue, 23 Jun 2020 15:32:37 +0000 (17:32 +0200)]
Fixes for building nGraph Python API (#707)
Anton Pankratv [Tue, 23 Jun 2020 14:23:47 +0000 (17:23 +0300)]
Hetero plugin supports ngraph (#530)
Egor Churaev [Tue, 23 Jun 2020 13:21:24 +0000 (06:21 -0700)]
[IE CLDNN] Fix device release with static plugin instance (#1034)
The problem was in order of freeing memory. _context was removed before
_device and it looks like cl::Device in destructor tries to read some
info from cl::Context. And in this case we got this problem with
addressing because the memory already was freed.
For fixing the problem I changed the order of constructing members. And
based on principle: "First constructed, last destructed", the problem
was fixed.
JIRA: 29649
Ilya Lavrenov [Tue, 23 Jun 2020 13:10:00 +0000 (16:10 +0300)]
Removed deprecated Data ctor (#1078)
Ilya Lavrenov [Tue, 23 Jun 2020 13:09:40 +0000 (16:09 +0300)]
Cnnnetwork deprecated methods (#1077)
* Removed getName with char *
* Removed getPrecision from ICNNNetwork
Andrew Bakalin [Tue, 23 Jun 2020 12:05:40 +0000 (15:05 +0300)]
[VPU] Support for originalLayersNames attribute in exec graph (#1033)
Andrew Bakalin [Tue, 23 Jun 2020 12:03:50 +0000 (15:03 +0300)]
[VPU] Fix eltwise broadcast (#1001)
* [VPU][Tests] Extend eltwise test cases
* [VPU] Fix Myriad2
* [VPU] Update firmware
* [VPU] Review fixes
* [VPU] Update old deprecated tests
Katarzyna Mitrus [Tue, 23 Jun 2020 11:11:46 +0000 (13:11 +0200)]
[ONNX importer] Add support for Usample-8 and Upsample-9 (#967)
Roman Kazantsev [Tue, 23 Jun 2020 10:49:14 +0000 (15:49 +0500)]
Correct removing nodes from graph and add test for ConstToResult transform (#1084)
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
Ilya Lavrenov [Tue, 23 Jun 2020 09:35:17 +0000 (12:35 +0300)]
CMAKE: cross-compilation for ia32 (#1074)
Irina Efode [Tue, 23 Jun 2020 07:05:28 +0000 (10:05 +0300)]
[IE TESTS] Add Squeeze single layer test (#985)
* [IE TESTS] Add Squeeze single layer test
* [IE TESTS] Add Unsqueeze test
* Fix some comments
Evgenya Stepyreva [Tue, 23 Jun 2020 06:58:03 +0000 (09:58 +0300)]
[ DTS ] MatMul (#974)
* [ DTS ] MatMul
* [ TESTS ] Dynamic MatMul inference test disabled
Konrad Dobros [Tue, 23 Jun 2020 06:40:15 +0000 (08:40 +0200)]
[IE CLDNN] Add work-around for 1d input to Gather (#1070)
Evgenya Stepyreva [Tue, 23 Jun 2020 05:27:27 +0000 (08:27 +0300)]
[ MO ] Hard-coded Interpolate followed by concat reshape-ability fixing (#818)
Ivan Tikhonov [Tue, 23 Jun 2020 05:22:12 +0000 (08:22 +0300)]
Fix for Kaldi models with a batch of more than 1 (#1012)
* Fix kaldi models (batch > 1)
* ngraph codestyle
* fix ngraph to ie conversion
* Added comment
* apply review comments
* Added test for the case using the SetBatchSize function when ReadValue op is in the network
* Check status code instead of message
* Use new ngraph api
Ivan Tikhonov [Mon, 22 Jun 2020 20:37:08 +0000 (23:37 +0300)]
Fix Android ARM build (#1032)
* fix arm build: size_t -> uint64_t
* apply static_cast
* fix dynamic_to_static_shape_binary_elementwise.cpp file
* add static_cast in prior_box.cpp
Maxim Vafin [Mon, 22 Jun 2020 17:42:02 +0000 (20:42 +0300)]
Cherry-pick fix OneHot transformation for Bert Squad opset 10 (#1068)
* Fix OneHot transformation for Bert Squad opset 10
* Add transformation for squeezing depth for OneHot
Denis Orlov [Mon, 22 Jun 2020 15:49:38 +0000 (18:49 +0300)]
[GNA] Initialize a local variable (#1067)
Pavel Rodionov [Mon, 22 Jun 2020 15:00:29 +0000 (18:00 +0300)]
[GNA] Support export model with multiple inputs/outputs and Permute layer (#775)
Vladimir Paramuzov [Mon, 22 Jun 2020 14:09:39 +0000 (17:09 +0300)]
[IE CLDNN] fsv4 to fsv16 conv (#1030)
Alexander Chaiko [Mon, 22 Jun 2020 14:02:23 +0000 (16:02 +0200)]
[IE CLDNN] WA to inconsistency between input and const 1d tensors for concat (#1063)
dmitrygo [Wed, 10 Jun 2020 07:37:28 +0000 (10:37 +0300)]
[CPU] Fixed issue with unsupported reorder case for groupped convolutions
Andrey Somsikov [Mon, 22 Jun 2020 13:23:09 +0000 (16:23 +0300)]
Add memcheck runner script (#1031)
Script executes measurement isolated with gtest-parallel,
handles database uploading and reports generation.
Anton Romanov [Mon, 22 Jun 2020 13:12:02 +0000 (16:12 +0300)]
Fix samples build script (#819)
Kamil Magierski [Mon, 22 Jun 2020 12:46:01 +0000 (14:46 +0200)]
Fix cases then const blob precision is not FP32/FP16 (#1020)
Co-authored-by: kmagiers <kmagiers@intel.com>
Adam Osewski [Mon, 22 Jun 2020 09:40:58 +0000 (11:40 +0200)]
[Py] Ngraph Py API TensorIterator (#718)
Evgeny Lazarev [Mon, 22 Jun 2020 08:52:00 +0000 (11:52 +0300)]
Cleanup IR v7 from the MO (#1008)
* Removed back phase transformations related to IRv7
* Fixed setting value for the input port using the 'set_value' method
* Removed front and middle phase transformations related to IRv7
* Cleanup the rest of the Model Optimizer transformations from IRv7 specific transformations
* Final cleanup of the deprecated IR v7 related code
* Removed 'blobs_as_input' usage in the Model Optimizer.
* Removed function '_fuse_add' from the Model Optimizer since it is not used anymore.
* Removed 'keep_in_IR' node attribute for FakeQuantize ops in the MO
* Disabled failing gpu_engine.user_context test
Ilya Lavrenov [Mon, 22 Jun 2020 08:35:44 +0000 (11:35 +0300)]
Remove some stuff from legacy library (#1043)
Edward Shogulin [Sat, 20 Jun 2020 16:06:26 +0000 (19:06 +0300)]
[LPT] BERT with specific biases support & improvement (cherry-pick to master) (#1021)
* [LPT] BERT with biases support
* [LPT] Gemm biases and quantization
* [CPU] Fixed FullyConnected + Depthwise node fusing
* [LPT] FullyConnected 3D: symmetric quantization support
* [LPT] FullyConnected 3D: symmetric quantization support fix
* [CPU] Fixed FullyConnected + Depthwise fusing initialization
Co-authored-by: dmitrygo <dmitry.gorokhov@intel.com>
Marcin Sielski [Fri, 19 Jun 2020 19:06:30 +0000 (21:06 +0200)]
Fix build issue (#923)
* Fix build issue
Why:
* Enable to build OpenVINO.
This change addresses the need by:
* Adding include directories,
* Removing IE::inference_engine_c_api dependency.
* Remove IE::inference_engine_nn_builder reference.
Why:
* Enable to build OpenVINO.
This change addresses the need by:
* Removing IE::inference_engine_nn_builder dependency.
Co-authored-by: Ilya Lavrenov <ilya.lavrenov@intel.com>
Konrad Dobros [Fri, 19 Jun 2020 18:41:08 +0000 (20:41 +0200)]
[IE CLDNN] Fix activation implementation for fsv16 format (#1037)
For b_fs_yx_fsv16 format in reference kernel features for dispatch are
rounded to multiple of 16. This change adds correct check in kernel to
return work-items that are inside this dispatch padding.
Previously those work-items could corrupt memory expected to be filled
with 0s, and for parametrized activation due to bounds checking with
modulo operator they could have been corrupting actual layer output.
Issue: CVS-27672
Ilya Lavrenov [Fri, 19 Jun 2020 18:04:12 +0000 (21:04 +0300)]
Moved plugin to hidden folder (#999)
Ilya Lavrenov [Fri, 19 Jun 2020 17:55:53 +0000 (20:55 +0300)]
Removed VPU option (#1027)
Vitaliy Urusovskij [Fri, 19 Jun 2020 12:56:32 +0000 (15:56 +0300)]
Add memcheck comparison script (#935)
Add compare_memcheck_2_runs.py compares two runs.
Add handling of broken files for `parse_memcheck_log`
Ilya Lavrenov [Fri, 19 Jun 2020 12:10:21 +0000 (15:10 +0300)]
Removed Int8 normalizer and statistics (#919)
* Removed Int8 normalizer and statistics
* Removed statistics handling from tests utils
* Fixed tests compilation with statistics
Andrey Sokolov [Fri, 19 Jun 2020 11:48:53 +0000 (14:48 +0300)]
[IE VPU] use optimized ReduceMean instead of GlobalPooling (#629)
Ilya Lavrenov [Fri, 19 Jun 2020 11:48:26 +0000 (14:48 +0300)]
Removed shape infer extension (#917)
Andrey Zaytsev [Fri, 19 Jun 2020 11:39:57 +0000 (14:39 +0300)]
Added opset docs (#992)
Maksim Doronin [Fri, 19 Jun 2020 10:22:31 +0000 (13:22 +0300)]
[IE VPU] Evaluate DSR (#770)
* [IE VPU] Add evaluate method to DSR
* [IE VPU] Enable DSR_Reshape tests
* [IE VPU] Improvements in DSR op
* [IE VPU] Fix typo in copyBlobAccordingUpperBound
* [IE VPU] Support dynamic inputs
* [IE VPU] Use dynamic inputs in tests
* [IE VPU] Improve conditions in propogateDynamism pass
* [IE VPU] Fix Myriad2 tests via dosabling reorder
* [IE VPU] make error message more explicit
* [IE VPU] Fix Win compilation: std::stoi in <string>
* [IE VPU] Improve data transferring to work with ND tensors
* [IE VPU] Avoid ODR in myriad common test utils
* [IE VPU] Split code in propagate dynamism into separate methods
* [IE VPU] Simplify conditions in DSR parsing
* [IE VPU] Emplace data in initialStages when remove stage order
Gorokhov Dmitriy [Fri, 19 Jun 2020 10:03:21 +0000 (13:03 +0300)]
[IE Common][WA] Skipped const folding for Convolution layer (#1005)
Andrey Dmitriev [Fri, 19 Jun 2020 07:21:04 +0000 (10:21 +0300)]
[GNA] fix permute 0_2_1 (#991)
Andrey Somsikov [Thu, 18 Jun 2020 22:27:32 +0000 (01:27 +0300)]
fix: inference-engine/ie_bridges/python/requirements.txt to reduce vulnerabilities (#1006)
The following vulnerabilities are fixed by pinning transitive dependencies:
- https://snyk.io/vuln/SNYK-PYTHON-NUMPY-73513
Co-authored-by: snyk-bot <snyk-bot@snyk.io>
Vladimir Paramuzov [Thu, 18 Jun 2020 16:55:17 +0000 (19:55 +0300)]
[IE CLDNN] Fix strided slice (#950)
Alexander Zhogov [Thu, 18 Jun 2020 16:20:03 +0000 (19:20 +0300)]
Azure CI: Add gtest-parallel on Lin & Mac (#980)
Jedrzej Hajduczenia [Thu, 18 Jun 2020 13:40:52 +0000 (15:40 +0200)]
[IE CLDNN] Add pooling b_fs_yx_fsv16 int8 (#565)
Mikhail Treskin [Thu, 18 Jun 2020 11:31:16 +0000 (14:31 +0300)]
Adding new layer tests to validation (#848)
* Add transpose, gather and reduce ops layer tests
* Fix skipping of Reduce Logical tests
* Fix compilation error with icl
Maxim Andronov [Thu, 18 Jun 2020 11:21:23 +0000 (14:21 +0300)]
[CPU] fix one dims scale shift (#989)
Nikolay Shchegolev [Thu, 18 Jun 2020 10:59:26 +0000 (13:59 +0300)]
[Common] Static analysed issues. Part II. (#881)
Nikita Kudriavtsev [Thu, 18 Jun 2020 10:49:20 +0000 (13:49 +0300)]
Changes: (#982)
- Named structures in bmp.h to avoid MSFT compiler error
- Fix for non-void function with missing return statement to avoid Intel compiler error
- Enabled "smoke_ExportUsingFileNameImportFromStreamNoThrowWithDeviceName" test
- Fix for MvncTest
Evgenya Stepyreva [Thu, 18 Jun 2020 08:36:07 +0000 (11:36 +0300)]
[ DTS ] Reduces (#940)
Evgenya Stepyreva [Wed, 17 Jun 2020 21:16:20 +0000 (00:16 +0300)]
[ DYN NMS ] Static & Dynamic ops; DTS transformation; VPU tests (#884)
Evgeny Lazarev [Wed, 17 Jun 2020 15:31:23 +0000 (18:31 +0300)]
Relaxed MO requirements for "protobuf" package (#864)
Co-authored-by: Evgeny Lazarev <elazarev.nnov@gmail.com>
Pavel Esir [Wed, 17 Jun 2020 15:27:37 +0000 (18:27 +0300)]
fixed some typos in MO help (#972)
Evgenya Stepyreva [Wed, 17 Jun 2020 15:20:08 +0000 (18:20 +0300)]
[ DTS ] Exp, Softmax, Greater (#926)
- Dynamic to Static transformation enabled for Exp, Softmax, Greater
- Logic Elementwises default ctor sets autobroadcasting according to spec
Ilya Churaev [Wed, 17 Jun 2020 14:17:32 +0000 (17:17 +0300)]
Fixed cpack information, removed some links (#976)
Konrad Dobros [Wed, 17 Jun 2020 12:07:18 +0000 (14:07 +0200)]
[IE CLDNN] Fix gather dimensions calculation (#959)
Ilya Lavrenov [Wed, 17 Jun 2020 11:42:41 +0000 (14:42 +0300)]
Execution graph via ngraph for CPU plugin (#510)
* Execution graph via ngraph for CPU plugin
* Fixes
* Migrated to VariantImpl instead of Parameter
* Reverted to dedicated ExecutionNode once again
* Re-use new execution graph in tests
* Fixed one more tests to use execution graph via ngraph::Function
Ilya Lavrenov [Wed, 17 Jun 2020 11:41:16 +0000 (14:41 +0300)]
CMAKE: fixed path for bin artifacts for 32bits (#890)
Ilya Lavrenov [Wed, 17 Jun 2020 11:34:38 +0000 (14:34 +0300)]
Pass SizeVector by const reference in ie_layout.hpp (#965)
Irina Efode [Wed, 17 Jun 2020 08:26:33 +0000 (11:26 +0300)]
[IE TESTS] disable Some myriad tests on Win (#763)
* [IE TESTS] disable Some myriad tests on Windisable Some myriad tests on Win
* Skip test with todo
Ilya Churaev [Tue, 16 Jun 2020 19:30:20 +0000 (22:30 +0300)]
Use creators only for default opsets (#948)
Gladilov, Gleb [Tue, 16 Jun 2020 13:17:36 +0000 (16:17 +0300)]
[IE][VPU]: Enables dynamic output from middle of network support (#930)
* [IE][VPU]: Enables dynamic output from middle of network support
This feature is very useful for debugging dynamic networks.
Changes include modification of existing addCopyForOutputsInsideNetwork
pass to respect dynamic outputs and moving propagateDynamismToOutputs
pass after addCopyForOutputsInsideNetwork. The motivation for last change
is to avoid unnecessary copy stages due to not synchronized logic, because
previously:
* First in Front-End (parseDSR) we mark shape data object as output
* Then in propagateDynamismToOutputs we insert copy stage for that case.
It's necessary if shape data object had other consumers
* Then in convertShapeNotation we insert Gather consumer for output data object
* Finally, addCopyForOutputsInsideNetwork inserts one more copy stage to leave
output data object without consumers.
Signed-off-by: Gladilov, Gleb <gleb.gladilov@intel.com>
* [IE][VPU]: Replaces attrs.has + attrs.get with attrs.getOrDefault
* [IE][VPU]: Fixes setting IE-notation and converted-notation to the same data object
Andrey Dmitriev [Tue, 16 Jun 2020 12:21:24 +0000 (15:21 +0300)]
[GNA] Added test ScaleShift and fixed power layer with non zero shift (#774)
* [GNA] Added test ScaleShift and fixed power layer with non zero shift
Ilya Lavrenov [Tue, 16 Jun 2020 12:06:48 +0000 (15:06 +0300)]
Removed deprecated error listener, getmappedtopology (#915)