IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5,864 Commits 38 Branches 54 Tags
2d73eb61573bde9884cd6c7ad520d70bc78a5430
Commit Graph

3428 Commits

Author SHA1 Message Date
Gorokhov Dmitriy
2d73eb6157 [CPU] Migration on oneDNN v2.4.4 (#9326) 
Co-authored-by: Anton Voronov <anton.voronov@intel.com>
 2021-12-21 14:45:44 +03:00
tgubanova-lohika
f9ece370f3 [GPU] Implement RandomUniform operation (#9043) 
* [GPU] Implement Random Uniform kernel

* [GPU] Add random uniform kernel selector

* [GPU] Add random uniform primitive

* [GPU] Add random uniform instantiation helpers

* [GPU] Implement unit test for the random uniform operation

* [GPU] Add random uniform operation builder in cldnn_engine plugin

* [GPU] Add single layer tests for the random uniform operation

* [GPU] Add conversion for Random Uniform output type

* [GPU] Add RTTI macroses to RandomUniform operation
 2021-12-21 13:24:07 +03:00
Aleksandr Korolev
4b0a2c9127 [VPU] Coverity fixes (#9214) 
Some coverity fixes
 2021-12-21 12:58:05 +03:00
Maxim Shevtsov
3032f20e03 brushed the #streams logic when #requests is limited (#9184) 2021-12-21 12:07:59 +03:00
Ivan Novoselov
fa9434494f [Snippets][CPU] Eltwise subgraphs code generation (#7936) 2021-12-20 22:45:01 +03:00
song, bell
1901f33bd3 auto dynamic shape (#8731) 
* initial commit

Signed-off-by: fishbell <bell.song@intel.com>

* overwrite GetBlob/SetBlob to support dynamic shape when needed

Signed-off-by: fishbell <bell.song@intel.com>

* extend workers

Signed-off-by: fishbell <bell.song@intel.com>

* clean up code

Signed-off-by: fishbell <bell.song@intel.com>

* fix lock and add case for CPU,GPU

Signed-off-by: fishbell <bell.song@intel.com>

* fix mem leak

Signed-off-by: fishbell <bell.song@intel.com>

* clean up allocated worker

Signed-off-by: fishbell <bell.song@intel.com>

* fix test case failure

Signed-off-by: fishbell <bell.song@intel.com>

* enable case for chaining

Signed-off-by: fishbell <bell.song@intel.com>

* fix arm compile issue

Signed-off-by: fishbell <bell.song@intel.com>

* clang format, fix potential threading issue

Signed-off-by: fishbell <bell.song@intel.com>

* clean up idle request queue if infer request destroyed

Signed-off-by: fishbell <bell.song@intel.com>

* keep multi legacy behavior

Signed-off-by: fishbell <bell.song@intel.com>

* elegant way to handle recycling

Signed-off-by: fishbell <bell.song@intel.com>

* avoid the exception with set output blob in auto dynamic case

Signed-off-by: fishbell <bell.song@intel.com>

* remove increase workers logic

Signed-off-by: fishbell <bell.song@intel.com>

* formatting

Signed-off-by: fishbell <bell.song@intel.com>

* remove the irrelevant code to dynamic shape

Signed-off-by: fishbell <bell.song@intel.com>

* filter target device to CPU in the case of dynamic shape

Signed-off-by: fishbell <bell.song@intel.com>

* fix logic in selectdevice, add case for GPU

Signed-off-by: fishbell <bell.song@intel.com>

* add case for AUTO:GPU

Signed-off-by: fishbell <bell.song@intel.com>

* fix build error

Signed-off-by: fishbell <bell.song@intel.com>

* add comment for skipped cases

Signed-off-by: fishbell <bell.song@intel.com>

* fix the case failure caused by naming conflict

Signed-off-by: fishbell <bell.song@intel.com>

* move file to new folder

Signed-off-by: fishbell <bell.song@intel.com>

* change to 2.0 API in plugin

Signed-off-by: fishbell <bell.song@intel.com>

* refactor mkldnn code change

Signed-off-by: fishbell <bell.song@intel.com>

* formatting

Signed-off-by: fishbell <bell.song@intel.com>
 2021-12-20 18:06:50 +03:00
Vladislav Golubev
ca7600edde [dynamism][Transformations] Sequences conversions: dynamic shapes support (#7766) 
* [Transformations] Sequences conversion: refactoring & dynamic shapes support

* tests moved
 2021-12-20 12:49:39 +03:00
Ilya Churaev
dbf8fcf556 Create ov_runtime library (#9199) 
* Create ov_runtime library

* Added dev target

* Fixed SOURCE_DIR property for interface lib

* Fixed missed include path

* Fixed SOURCES for plugins and frontends

* Fixed Windows export

* Try to fix preproc build

* Fixed MKLDNNPlugin

* Fix C API

* Try to fix public CI

* Fixed comments

* Use TARGET_NAME variable

* Fixed static build

* Fixed target name in comfig

* Revert "Fixed target name in comfig"

This reverts commit f61c1e0c09.

* Try to remove WA with include path

* Try to fix TBB dependency for old CMake

* Revert install to lib for legacy tests

* Try to fix public tests

* Add WA for v7 reader

* Fixed dev export name

Co-authored-by: Ilya Lavrenov <ilya.lavrenov@intel.com>
 2021-12-20 11:59:08 +03:00
Edward Shogulin
37ad512d98 [CPU] Optimize quantization scheme for SPR/ATS (#7549) 
* [CPU] Optimize quantization scheme for SPR/ATS

* [CPU] [LPT] plugin tests

* [GPU] [LPT] plugin tests

* [CPU] limitation was removed

* [CPU] optimization FP32 old way support

* [LPT] comment fix

* [LPT] Multiply plugin test improvement

* [LPT] Multiply support

* [LPT] GPU tests fix

* [LPT] test quick fix

* [LPT] new ppi fix

* look like spent time for tests refactoring
 2021-12-20 11:19:40 +03:00
Egor Duplensky
abee3ea4d4 [CPU] Refactoring. Avoid using align arg when appending post ops (#9225) 
Always align legacy scale shift post ops
 2021-12-20 10:23:32 +03:00
Jade Cho
a4518ae595 [GPU] Add DumpLayersLimitBatch config to GPU debug utils. (#9196) 
* Add DumpLayersLimitBatch config to GPU debug utils.

+ Support OV_GPU_Help config
+ Only run first inference if OV_GPU_DumpLayersPath is set.
+ Fix dump graph bug.

* Apply some comments

* Remove unnecessary code.
 2021-12-20 14:19:53 +09:00
Egor Duplensky
b394441c79 [CPU] Fix sse4 uni_ methods (Normalize and Input nodes) (#9268) 2021-12-17 15:13:16 +03:00
Sergey Shlyapnikov
71f3e2da10 [GPU] Fix fused ops reshaping for OneDNN FC (#9217) 2021-12-17 13:26:59 +03:00
Yaroslav Torzuk
3f5fca80db [GPU] Slice operation (#8267) 2021-12-17 12:41:05 +03:00
Aleksandr Korolev
4c4d006c5a [VPU] restore network P/R info in Import method (#8844) 
In the new API, we need to recover information about function inputs/outputs.
 2021-12-17 12:14:08 +03:00
Maxim Andronov
04dc16f861 [CPU] General fixes for dynamic shapes. Part 2 (#8871) 2021-12-17 11:12:07 +03:00
Min, Byungil
94c3731b50 [GPU] support multiple sum post-op of onednn kernel (#9221) 
+ Applied WA for multiple sum post-ops of nednn kernel

Signed-off-by: Min, Byungil <byungil.min@intel.com>
 2021-12-17 10:43:21 +03:00
Sergey Lyubimtsev
b41cb57773 Fix build issue for openvino wheel package on Windows (#9265) 
* Fix build issue for openvino wheel package on Windows

* fix env
 2021-12-17 10:42:21 +03:00
Yury Gaydaychuk
58f0c75b91 [CPU] RoiAlign: case of integer tensor handled (#8069) 2021-12-16 22:02:09 +03:00
Vladislav Volkov
6ddc47a7ef [CPU] Convert precisions on inputs/outputs (#8805) 2021-12-16 16:58:19 +03:00
Sergey Shlyapnikov
2514c0ef38 [GPU] Add gemm_tiled_opt i8/u8 output support (#9202) 2021-12-16 15:20:28 +03:00
Alexander Zhogov
d10e8005c0 Revert "Fix build issue for openvino wheel package on Windows (#9231)" (#9260) 
This reverts commit 460a6634fd.
 2021-12-16 14:33:56 +03:00
Sergey Lyubimtsev
460a6634fd Fix build issue for openvino wheel package on Windows (#9231) 
* fix build issue for openvino wheel package on Windows

* revert ngraph_libs
 2021-12-16 13:01:03 +03:00
Roman Lyamin
2068d5838b [GPU] Add I420toRGB/I420toBGR operations (#9204) 2021-12-16 10:27:27 +03:00
Alexandra Sidorova
58be795970 [CPU] Added BF16 support for CumSum as well (#9030) 2021-12-16 10:24:57 +03:00
Maxim Andronov
d5f84ad783 [CPU] Deconvolution dynamism support (#8512) 2021-12-16 09:53:14 +03:00
Ilya Churaev
9aedece398 Fixed cpu tests location (#9224) 2021-12-15 14:32:43 +03:00
Sergey Shlyapnikov
1177d2b282 [GPU] Change FQ output for first Convolution (#9200) 
* update onednn_gpu

* [GPU] Add bs_fs_yx_bsv8_fsv4 format

Co-authored-by: Kim,SungEun <sungeun.kim@intel.com>
 2021-12-15 13:15:13 +03:00
Ilya Znamenskiy
b492b59136 [GPU] Fix for conv/deconv weights calculated in runtime (#8952) 2021-12-15 12:17:13 +03:00
Zhang Yi
4fba88d29a Yi3/shape inference 3rd batch (#8611) 2021-12-15 10:23:19 +03:00
Ilya Churaev
e8d5cf43d0 Move ie tests (#9161) 
* Moved inference engine tests

* Fixed cmake

* Fixed paths

* Try to fix build

* Try to fix fluid tests

* Try to fix CI

* One more fix

* New issue

* Additional fix

* Disable Wall for tests

* Fixed old unit tests

* Revert "Disable Wall for tests"

This reverts commit 8b6e058f59.

* Fixed some tests

* Fixed master build

* Fix build

* Fixed some warning

* Fixed merge conflicts

* Try to fix LTO
 2021-12-15 08:50:56 +03:00
Egor Duplensky
3f6a026ae9 [CPU] Enable matmul deconv bin postops (#8009) 2021-12-14 19:44:38 +03:00
Irina Efode
2b9c4a7f42 [IE TESTS] Fix run of behavior GPU plugin config tests (#9178) 
* [IE TESTS] Fix run of behavior GPU plugin config tests

* Return extra removed

* remove extra lines
 2021-12-14 18:09:34 +03:00
Sungeun Kim
bf3046e3a0 [GPU] Integration Prelu fusing (#8958) 
* [GPU] Add new data format: os_is_yx_osa2_isa8_osv8_isv2

* fusing prelu

- add onednn_post_op_type::binary_relu

* update onednn_gpu
 2021-12-14 20:09:50 +09:00
Roman Lyamin
7d0d0ea503 [GPU] Add NV12toRGB/NV12toBGR operations (#8838) 2021-12-14 12:01:50 +03:00
Andrey Noskov
e06d9fdbb9 [GNA] Added params to LSTMCell (#8714) 
* [GNA] Added params to LSTMCell
   - parametrized num of Cells

* Fixed review comments
 2021-12-14 11:03:26 +03:00
hyunback kim
d09bbb498e [GPU] Support usm_device dump. (#9103) 
* [GPU] Support usm_device dump.

When lock/unlock copy to host buffer from device.

Signed-off-by: hyunback <hyunback.kim@intel.com>

* Remove dependency check in dump with usm.

Signed-off-by: hyunback <hyunback.kim@intel.com>

* Apply codereview feedback.

Signed-off-by: hyunback <hyunback.kim@intel.com>
 2021-12-14 14:58:52 +09:00
Irina Efode
7335387db9 [IE TESTS] Update SubgraphBase class (#9148) 2021-12-14 01:09:40 +03:00
Anton Pankratov
ef58ec6c8c Added get compiled model API (#9186) 2021-12-14 00:14:01 +03:00
Anastasia Popova
c39dba62b0 Partial shape serialization and deserialization in nGraph (#8859) 
* Added support of partial shape serialization and deserialization.

* Small correction.

* Conflicts resolve.

* Clang-format.

* Fixed checks.

* Added partial shape, dimension support in visitor.

* Code style.

* Code style.

* Fixed IR reader parameter extender for partial shape.

* Moved IR reader changes to MO PR.

* Added rank check.

* Fixed dimension test.

* Code refactoring.

* Undo refactor.

* Code style

* Removed unnecessary code changes from cnn network.

* Extended visitor tests for PartialShape.

* Code style

* Removed code duplications, added dynamic rank test.

* Code style.

* Added negative tests for deserialization.

* Align with changes on master.

* Changed ov::Function to ov::Model.

* Fixed tests.

* Small fix.

* Fixed tests.

* Code style.
 2021-12-14 00:09:45 +03:00
Ilya Znamenskiy
06865a252a [GPU] TopK optimizations for value mode and OPERATION_NUM = 1 (#9170) 
[GPU] Cycle unrolling and minimization of unused thread number
 2021-12-13 19:11:18 +03:00
Anastasia Popova
5124d95df7 Old API map element type applying fix (#8735) 
* Changed logic of applying old_api_map_element_type.

* Code style.

* Added validate_nodes_and_infer_types().

* Fixed MeanScale for uint8 parameters.

* Added checks.

* Small fix.

* Added unit test.

* Merge conflicts fix.

* Small correction.

* Small corrections.

* Added type correction.

* Added comment.

* Fix conflicts.

* Test fix.

* Small fix.

* Fixed conflicts.
 2021-12-13 14:51:09 +03:00
Edward Shogulin
d109cf3348 [LPT] ONNX support extension: INT8 zero-point + FQ folding fix (#9052) 
* [LPT] ONNX support extention: INT8 zero-point + FQ folding fix

* [LPT] ReduceMean plugin tests extension
 2021-12-13 14:19:31 +03:00
Ilya Lavrenov
11f0ec8f46 Legacy as STATIC (#5691) 
* Legacy library is static

* Don't export symbols from legacy

* LEgacy only for cnn_network_ngraph_impl.cpp

* Removed Windows specific functions calls

* Use static_cast

* Fixed Windows

* Removed legacy library to DM configs
 2021-12-13 13:02:53 +03:00
Mang Guo
a21241ae0d Mang/shape inference interpolateV4/interpolateV0/ScatterElementUpdate/ScatterNDUpdate (#8521) 2021-12-13 12:26:03 +03:00
Irina Efode
a64dc2d238 [IE TESTS] Fix conformance mode in report according to updated XML files (#9147) 2021-12-13 12:08:04 +03:00
Zhang Yi
94195c7aba Yi3/shape infer 1st batch (#9105) 2021-12-13 11:32:31 +03:00
Paul Youngsoo Ahn
4e8a6d5a4b [GPU] Add sub/div mode in eltwise fusing (#8759) (#8759) 
- Add sub/div operation of eltwise fusing
- Propergate dep ordered info to Jitter
- Add sub/div to the fusing gpu tests
 2021-12-13 10:24:37 +03:00
Ilya Churaev
4f62a48077 Remove ngraph backends (#9162) 2021-12-13 00:04:56 +03:00
Ilya Lavrenov
9e519946f0 Integrate JSON libs (#9145) 
* Add nlohmann json (Release 3.10.4) as submodule

* Move nlohmann_json lib to json folder, add json_schema validator lib as submodule

* Move BUILD_SHARED_LIBS flag to a separete scope

* Add export of nlohmann_json_schema_validator

* Fix build

* set folder thirdparty

* link lib to offline_transformations and benchmark_app

* suppress shadowing names warning in nlohmann_json lib

* fix include in benchmark_app

* Resolve review comments: add json subdirs to samples cmake

* Fix static build

* Proper json integration

* removed cpp_samples_deps component

Co-authored-by: Ivan Tikhonov <ivan.tikhonov@intel.com>
 2021-12-12 20:40:41 +03:00