Commit Graph

3280 Commits

Author SHA1 Message Date
Sergey Lyubimtsev
6d28be7411
Enable CPack for Python samples (#5406)
* Enable Python samples for CI & CPack

* remove external dependency for CMakeLists file
2021-04-27 18:53:55 +03:00
Ilya Lavrenov
6c83e0f8a5
Small refactoring in TEMPLATE plugin (#5398)
* Small refactoring in TEMPLATE plugin

* Fixed compilation on Windows

* Fixed code style
2021-04-27 18:52:45 +03:00
Gleb Kazantaev
689f8aedb6
Added BackwardGraphRewrite And Input<Node> RTInfo (#5343)
* Added BackwardGraphRewrite

* Add RT Info to Input<Node> class

* Add RTInfo tests; BackwardGraphRewrite tests
2021-04-27 17:26:20 +03:00
Svetlana Dolinina
3028c78594
fix typo + test (#5392) 2021-04-27 13:37:02 +03:00
Ilya Lavrenov
fa7d67b07f
Removed reading of port precision (#5397) 2021-04-27 10:18:19 +03:00
Ilya Lavrenov
0337b055db
Added GetAvailableDevices for AUTO device (#5377)
* Added GetAvailableDevices for AUTO device

* Fixed tests compilation
2021-04-27 09:12:55 +03:00
Mikhail Nosov
05cae2714f
Generate unique file names for GNAAOTTests (#5403)
Issue can be reproduced on stress testing:
Get stress scripts from git@github.com:google/gtest-parallel.git
> python3 <workdir>/gtest-parallel/gtest_parallel.py  ./InferenceEngineUnitTests --gtest_filter=*GNAAOT* -r 100

It starts each test separately in different threads/processes and there is big chance of conflicts with unit_tests.bin
2021-04-27 08:59:20 +03:00
Patryk Elszkowski
5de5f4d7d1
Constant op SLT (#5349)
* add SLT for Constant OP

* add test for U4 and I4

* drop test for BIN data

Co-authored-by: Patryk Elszkowski <patryk.elszkowki@intel.com>
2021-04-27 07:09:23 +03:00
Mikhail Nosov
6581127114
[Caching] Don't propagate import error on plugin side (#5390)
* HotFix for ConvertPrecision tests

* [Caching] Don't propagate import error on plugin side

Instead, invalid cache will be silently removed and network will be loaded without cache

Co-authored-by: Ilya Churaev <ilya.churaev@intel.com>
2021-04-27 06:54:49 +03:00
Ilya Churaev
2a51a62d48
Fixed Ninja build (#5408) 2021-04-27 06:36:45 +03:00
Artyom Anokhov
16da65fc93
IE PyAPI CMakeLists.txt: Added copying requirements.txt to <INSTALLROOT>/python folder (#5407) 2021-04-26 22:56:25 +03:00
Alexandra Sidorova
5ca7cba451
[CPU] Added improvements for Mish (#5062) 2021-04-26 18:11:41 +03:00
Sergey Lyubimtsev
5912c983d4
Enable make install for openvino/tools folder (#5326)
* enable make install for openvino/tools folder

* fix component name

* use python_tools as component name

* update ie_cpack_add_component name

* enable CPack for python tools

* use find_package(PythonInterp)
2021-04-26 18:06:39 +03:00
Tomasz Dołbniak
e15d9efe30
Template plugin support in ONNX compliance tests (#5375) 2021-04-26 15:34:39 +02:00
Vladislav Golubev
32ac4f4e15
[LPT] Reduce*Transformation (#4713)
* [LPT] added notAllChildrensAreFQ to NetworkHelper

* [LPT] Added ReduceBaseTransformation class

* [LPT] ReduceMeanTransformation

* [LPT][TESTS] ReduceMeanTransformation: functional and plugin tests

* [LPT] ReduceMaxTransformation

* [LPT][TESTS] ReduceMaxTransformation: functional and plugin tests

* [LPT] ReduceMinTransformation

* [LPT][TESTS] ReduceMinTransformation: functional and plugin tests

* [LPT] RedueSumTransformation

* [LPT][TESTS] ReduceSumTransformation: functional and plugin tests

* [nGraph] TypeRelaxed::clone_with_new_inputs: added validate_and_infer_types()

* [LPT] ReduceMin/ReduceMax: handle mul with negative values

* [LPT] ReduceBaseTransformation:
    * fold for dequantization constants instead of manual manipulations
    * refactoring

* [TESTS] Reduce*Transformation template functional test-class and builder

* codestyle fix

* [LPT] ReduceMeanTransformation: removed WA

* [LPT] added description for ReduceBaseTransformation

* [LPT] NetworkHelper::foldDequantizationConstant quick fix

* [LPT][TESTS] reduce builder quick fix and fp16 tests for gpu enabled
2021-04-26 16:03:03 +03:00
Ilya Churaev
d150d75a65
HotFix for ConvertPrecision tests (#5391) 2021-04-26 15:48:55 +03:00
Roman Lyamin
16d7d01a27
[IE CLDNN] Eltwise b_fs_zyx_fsv16 optimization (#5098) 2021-04-26 15:36:38 +03:00
Maksim Kutakov
9f2a3d0edc
[CPU] Split supported PD initialization refactoring plus tests for multiply port connections. (#4521) 2021-04-26 15:18:26 +03:00
Roman Donchenko
ed5313b2e1
Fix common misspelling: wraper -> wrapper (#5355) 2021-04-26 13:42:12 +03:00
Ilya Churaev
9045c94e3d
Improved performance of i4 conversion (#5258) 2021-04-26 13:12:35 +03:00
Irina Efode
7226190b30
[IE TESTS] Fix typo in script (#5384) 2021-04-26 12:54:33 +03:00
Egor Duplensky
93f63ceb6a
[CPU] Exclude colon characters from test names (#5243)
* [CPU] Exclude colon characters from test names

":" is used for specifying list of filters in '--gtest_filter=' option
So when a test name contains a colon char it is impossible to run such
test by its full name.
There is now way to escape this colon char as well.
2021-04-26 12:30:02 +03:00
Szymon Irzabek
af6f315071
[GNA] Remove FQ layers positioned after a memory layer (#5323) 2021-04-26 11:44:18 +03:00
Irina Efode
d2868c9873
[IE TESTS] Update conformance Readme file (#5374)
* [IE TESTS] Update conformance Readme file

* [IE TESTS] Update conformance Readme file
2021-04-26 11:17:31 +03:00
Mikhail Nosov
e53249ddf2
Disable time measurement (#5376)
On CI machines there is a chance to run various test executables at the same time.
When intensive CPU load occurs, 'sleeps' can last much longer than expected even in very pessimistic scenario
2021-04-26 10:28:26 +03:00
Kate Generalova
add383c566
feat: clang-style for cython api (#5354) 2021-04-26 09:40:20 +03:00
Egor Duplensky
39e1a21c42
[CPU] Enable bf16 RNN primitives (#4942) 2021-04-25 22:18:38 +03:00
Kate Generalova
8bb73273f1
fix: string comparing in object_detection_sample_ssd_c (#5371) 2021-04-25 21:42:04 +03:00
Ivan Novoselov
594c4882e6
[CPU] optimize ExtractImagePatches (#4898) 2021-04-25 19:52:07 +03:00
Irina Efode
b770be227b
[IE TEST] fix small issue in summarize (#5367)
* [IE TEST] fix small issue in summarize

* Add
2021-04-24 12:29:46 +03:00
Roman Kazantsev
fcea3f8a0c
Implement nGraph shell for Einsum-7 (#5282)
* Implement nGraph shell for Einsum-7

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

* Correct doxygen formats

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

* Apply clang format change

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

* Support implicit mode and capital letters

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

* Correct and optimize the code based on review

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

* Correct private methods and its API, add more tests

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

* Make equation aux methods public and remove regex usage

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

* Make is_subscript_correct function local

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

* Correct check for missed ellipsis and add test for it

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
2021-04-24 09:41:48 +03:00
Steve Yoo
7d0cae8bb5
Add test cases for PReLU in cpu plugin (#5293)
* For case when slope is vector
2021-04-23 17:19:35 +03:00
Kate Generalova
2063f17391
fix: doxygen links for samples (#5365) 2021-04-23 16:40:08 +03:00
Mateusz Bencer
9a569805c2
remove redundant header (#5363) 2021-04-23 16:35:46 +03:00
Ilya Lavrenov
0a8c3ef5a9
Fixed python test_utils compilation for clang-12 (#5362) 2021-04-23 16:34:26 +03:00
Gabriele Galiero Casay
59a001580f
Revise Split operation reference implementation (#5174)
* Review Split operation class

   * Added node validation checks
   * Added type_prop unit tests with invalid cases and negative axis value

* Add ngraph check for host tensor inputs and outputs

* Enable CPU backend tests

* Add serialization single layer tests

* Add more precisions, axis and num of splits in single layer tests

* Add op to script of trusted operations

* Fix check macro for inputs and outputs in evaluate method

* Fix comparison of integer expressions of different signedness compilation error

* Add validation check for upper bound dimension for dynamic shapes

* Add node validation check for num_splits attribute greater than zero

* Align spec with num_splits attribute range of values
2021-04-23 15:58:08 +03:00
Mikhail Ryzhov
8d2bc2b895
[PyPI] Corrected OS support matrix (#4919)
+ removed optional --system-site-packages from the instructions
2021-04-23 15:01:59 +03:00
Vladislav Golubev
05c23dfd94
[LPT][CommonOptimizations] ShuffleChannelsTransformation and ShuffleChannelsFusion (#4916)
* [nGraph] op::v0::ShuffleChannels: added type_info macros

* [LPT] ShuffleChannelsTransformation

* [CPU] ShuffleChannels decomposition disabled

* [LPT][TESTS] ShuffleChannelsTransformation functional tests

* [LPT][TESTS] ShuffleChannelsTransformation plugin tests

* [CommonOptimizations] Added ShuffleChannelsFusion transformation

* [CommonOptimizations][TESTS] ShuffleChannelsFusion tests

* refactoring and adding comments

* [CommonOptimizations] ShuffleChannelsFusion refactoring and fixes

* [CommonOptimizations] ShuffleChannelsFusion: removed unnecessary check

* [CommonOptimizations] transformation refactored and test-cases with dynamic shape added
2021-04-23 13:26:53 +03:00
Anton Chetverikov
576e692b1d
Update errors raising cases in modules versions checker in MO (#5141)
* Change error raising to message printing

* Update modules versions check

* Update check and add unit test

* Fix wrong function call

* Update unit tests

* Fix test

Co-authored-by: achetver <anton.chetverikov@.intel.com>
2021-04-23 13:23:06 +03:00
Victor Kuznetsov
e121fbd7de
Add int8 models to memcheck tests (#5333) 2021-04-23 12:56:59 +03:00
Chenhu Wang
1426dc7d68
[CPU] Activations are fused with MVN (#5272) 2021-04-23 11:38:18 +03:00
Chenhu Wang
528d4b1bac
contain only some typical fusion, reduce tensor size, not skip on traget without avx512 (#5361) 2021-04-23 11:29:53 +03:00
Kate Generalova
6510a68b10
fix: exclude .clang-format files from drop (#5353) 2021-04-23 10:07:40 +03:00
Ilya Lavrenov
4073e16a9e
Fixed compilation for deprecated API (#5356)
* Fixed compilation of offline API

* Fixed compilation with gcc 5.4

* Fixed compilation of ONNX importer dependencies
2021-04-23 09:27:39 +03:00
Patryk Elszkowski
576e750317
Constant op class update for Low Precision (#5311)
* add support for u1, u4 and i4 in op::v1::Constant

* fix convert_i4_to_string test

* Fix makeBinaryConvolution to fill memory allocated for LP types

* Add unit-test for range validation

Co-authored-by: Patryk Elszkowski <patryk.elszkowki@intel.com>
2021-04-23 07:31:04 +03:00
Patryk Elszkowski
6a850b1e7b
refactor convertOutputPrecision (#5338)
Co-authored-by: Patryk Elszkowski <patryk.elszkowki@intel.com>
2021-04-23 07:07:23 +03:00
Vladimir Paramuzov
cf6acfde78
[IE CLDNN] Fixed ambiguous native_powr call in mvn for fp16 out type (#5298) 2021-04-22 19:37:45 +03:00
Gleb Kazantaev
28cad9e3fb
Add transformation pipline for POT (#5328) 2021-04-22 19:13:14 +03:00
Patryk Elszkowski
cc7ae5e6d1
add for serialization names for i4 u4 (#5337)
Co-authored-by: Patryk Elszkowski <patryk.elszkowki@intel.com>
2021-04-22 18:41:41 +03:00
Ivan Kochin
636736825c
OpenVINO Hybrid CPUs support internal interface (#4602)
* OpenVINO Hybrid CPUs support

* Remove custom::task_arena abstraction layout

* Get back to the custom::task_arena interface

* Add windows.h inclusion

* Fix typo in macro name

* Separate TBB and TBBbind packages

* Fix compile-time conditions

* Fix preprocessors conditions

* Fix typo

* Fix linking

* make linking private

* Fix typo

* Fix target_compile_definitions syntax

* Implement CMake install logic, update sha hash for the tbbbind_2_4 package

* Add tbbbind_2_4 required paths to setup_vars

* Update CI paths

* Include ie_parallel.hpp to ie_system_conf.cpp

* Try to update dependencies scripts

* Try to fix dependencies.bat

* Modify dependencies script

* Use static tbbbind_2_4 library

* Remove redundant paths from CI

* Update tbbbind package version

* Make custom::task_arena inherited from tbb::task_arena

* Incapsulate all TBB versions related logic inside the custom namespace

* Move custom layer header to internal scope + minor improvements

* Introduce new ThreadBindingType + fix compilation

* Fix compilation

* Use public tbbbind_2_4 package

* Apply review comments

* Fix compilation without tbbbind_2_4

* Fix compilation with different TBB versions

Co-authored-by: Kochin, Ivan <ivan.kochin@intel.com>
2021-04-22 18:11:13 +03:00