IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
7,834 Commits 38 Branches 54 Tags
b149ea0e42e047e1e12a15fe2075e10008da9afd
Commit Graph

647 Commits

Author SHA1 Message Date
yanlan song
72c3bf222b fix coredump when quit benchmark_app (#13026) 
* fix coredump when quit benchmark_app

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

* enable tests

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

* add macro to handle CPU not built

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

Signed-off-by: fishbell <bell.song@intel.com>
 2022-09-15 16:47:11 +08:00
Evgenya Stepyreva
af16ea1d79 Revert "Fix experimental detectron do ref impl (#10621)" (#12683) (#13009) 
* Revert "Fix experimental detectron do ref impl (#10621)"

This reverts commit d87233863d.

* Disabled Experimental Detectron per agreement with GPU team. Ticket to fix it: 90209
 2022-09-12 18:16:13 +04:00
Sergey Shlyapnikov
af29d221b4 [GPU] Add NV12 -> Grayscale mode support (#12988) 
* [GPU] Add NV12 -> Grayscale mode support

* Fix uv plane shape
 2022-09-09 19:00:37 +04:00
yanlan song
facf990dfd fix inconsistent tbb config due to executor used in multi (#12929) 
* fix inconsistent tbb config due to executor used in multi

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

* refine comment

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

Signed-off-by: fishbell <bell.song@intel.com>
 2022-09-08 13:34:22 +08:00
Mateusz Tabaka
41fa6f360b Explicitly link onednn with tbb for tbb version in [2018,2019.4] (#12789) (#12837) 
Ticket: 89800
 2022-08-31 17:14:54 +03:00
Gorokhov Dmitriy
a0b661a274 [CPU] Fixed MHA accuracy for mixed precision case (#12820) 2022-08-31 10:53:38 +04:00
Tomasz Dołbniak
72d7b518ca cltools update to 22.08 [2022/2] (#12690) 
* cltools update to 22.08

* Hash update

* Hash update

* Adjustments for the new package
 2022-08-26 15:28:40 +04:00
Sergey Shlyapnikov
41a404f290 [GPU] fix Transpose issue for ConvertColor with FakeQuantize. (#12645) (#12761) 
Co-authored-by: Tang Wei <wei1.tang@intel.com>
Co-authored-by: Kurt Chen <kurt.chen@intel.com>
 2022-08-26 12:29:21 +04:00
Sergey Shlyapnikov
429c7265df [GPU] Implement NMS-9 operation (#11890) (#12760) 
* Fix GPU NonMaxSuppression implementation

* Introduce Nms9 single layer tests

* Adapt internal NMS and GPU implementation for NMS9 implementation

* Adapt CPU implementation in GPU for NMS9

* Add blocked layouts support to NMS

* Add unit tests for blocked formats for NMS

* Fix boxes groups size for the small shapes

* Use ocl implementation for blocked layout input

* Fix templates typedefs to pass win build

* Fix second output to set data in correct format

Co-authored-by: Tetiana Gubanova <tgubanova@lohika.com>
 2022-08-26 00:37:20 +04:00
Sergey Shlyapnikov
a3f8cef198 [GPU] Shared memory optimization for network::execute_impl() call (#12748) 2022-08-25 15:49:56 +04:00
guozhong wang
f409e95768 do not remove cpu when bind buffer (#12556) 
Co-authored-by: Shen, Wanglei <wanglei.shen@intel.com>
 2022-08-25 09:05:42 +03:00
Chen Xu
1e5fec7e25 [CPU] Reduce node improve performance for nspc layout (#12671) 2022-08-24 15:39:55 +04:00
Luwei Zhou
aa1a607328 [CPU] Fix the strided slice issue when ellipsis_mask has redundant data. (#12705) 2022-08-24 09:43:08 +04:00
Andrei Kochin
f87e00398d updated to convert b_fs_yx_fsv16 to o_is_yx_isv16 (#12630) (#12675) 
Co-authored-by: Eddy Kim <eddy.kim@intel.com>
 2022-08-23 15:46:54 +03:00
Gorokhov Dmitriy
a6bfc0cf0e [CPU] Support MHA optimization (#12643) 
* [CPU] Support MHA optimization

* [CPU] Extend pattern supported by MHA node

* [CPU] MHA: fixed int8 perf issue

Co-authored-by: Gu, Jianan <jianan.gu@intel.com>
 2022-08-23 12:50:02 +04:00
yanlan song
4d9443eb0e do not call get_profiling in threads (#12635) 
* do not call get_profiling in threads

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

* indent

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

Signed-off-by: fishbell <bell.song@intel.com>
Co-authored-by: Chen Peter <peter.chen@intel.com>
 2022-08-23 13:50:52 +08:00
Luo Cheng
e03fbd5c15 [CPU] Default enable avx512 f32 brgconv (#12620) 2022-08-19 17:59:15 +04:00
Ilya Lavrenov
29628a89b7 Tbb port (#12541) 
* Fixes for TBB 2018-2019.4

* Fixed CVS-89248
 2022-08-15 06:26:47 +04:00
Mateusz Tabaka
c0212a361a [CPU] Add RDFT and IRDFT operators (#12290) 
Tickets: 79178 and 79192

Co-authored-by: Mateusz Bencer <mateusz.bencer@intel.com>
 2022-08-12 14:10:53 +02:00
Mateusz Bencer
e628fae196 [GPU] Decompose NormalizeL2 for not supported cases (#12404) 2022-08-11 11:32:03 +02:00
Min, Byungil
f0f6896fc0 [GPU] Fix network loading time related to onednn engine creation (#12492) 
+ benchmark cache_dir option takes longer than cl_cache_dir env in loading network.
+ For clDNN execution, benchmark cache_dir created onednn_engine if just ONEDNN_ENABLE config is ON.
+ Creation of onednn_engine in ocl_engine is changed to on-demand.

Signed-off-by: Min, Byungil <byungil.min@intel.com>

Signed-off-by: Min, Byungil <byungil.min@intel.com>
 2022-08-11 09:32:20 +04:00
River Li
d328b00e48 [CC]Fix CC issue for transformation (#12292) (#12489) 
* Revert "Fixed 3 naming issue"

This reverts commit a92d3cfff5.

* Revert "Fix CC issues for transformation and snippets"

This reverts commit d08a3f5aac.

* Fix NGRAPH_PASS_CALLBACK issue to make it can work

* Fix matcher name missing issue
 2022-08-10 11:36:51 +04:00
Wilson Seok
1788c86943 change to node.weights() from weights_memory(0) (#12407) 2022-08-10 16:18:58 +09:00
Andrew Kwangwoong Park
ea302afb47 Update pre_replace_deconv to support output_shape for transposed conv (#12418) 
Signed-off-by: Andrew Park <andrew.park@intel.com>
 2022-08-10 10:37:51 +09:00
Ilya Churaev
c9f9795d29 Fixed newAPI for case if core was removed (#12208) 
* Fixed newAPI for case if core was removed

* Fixed code style

* Fixed typo

* Use new API by default

* Create core with template plugin

* Added doxygen comment

Co-authored-by: Ilya Lavrenov <ilya.lavrenov@intel.com>
 2022-07-23 11:53:26 +00:00
Kelvin Choi
3a72200f92 [GPU] Add reorder from i32 to f32 for max-pooling/conv/fc which doesn't support i32 (#12144) 2022-07-20 22:14:22 +09:00
Egor Duplenskii
fdae95a769 [CPU] Explicitly enable DNNL_VERBOSE only in case of CPU_DEBUG_CAPS (#12151) 
and rely on oneDNN default behavior otherwise
 2022-07-20 14:07:42 +04:00
Chenhu Wang
123f8e62bf [DOC][CPU] Denormals optimization document (#12132) 2022-07-18 16:37:44 +04:00
Taylor Yeonbok Lee
8c80f9ff58 [GPU] optimize permute_ref (#12160) 
* change memory access pattern of fsv layout for permute

* Fix permute_ref to process F first only when (bf...) => (b...f)

* Refactor

Co-authored-by: si-eun-kim <sieun.kim@intel.com>
 2022-07-18 18:26:00 +09:00
Eddy Kim
de5e9bb397 Revert "[GPU] Pass activation unit tests on DG2 (#11969)" (#12165) 
This reverts commit 3334e8933c.
 2022-07-18 18:25:45 +09:00
zihan wu
32f800c6a6 [CPU] polish onednn cc readme (#12114) (#12176) 2022-07-15 16:36:31 +00:00
Min, Byungil
b492f98d30 [GPU] modify fusing condition for reduce (#12147) 
Signed-off-by: Min, Byungil <byungil.min@intel.com>
 2022-07-15 16:07:43 +09:00
Andrew Kwangwoong Park
9c49b71c11 Enable tensor offset to GemmKernelRef for input padding support (#12140) 
Signed-off-by: Andrew Park <andrew.park@intel.com>
 2022-07-15 16:01:35 +09:00
Luo Cheng
4412e1ddfa [CPU] revert pr 11990 and enable brgconv avx512 on SPR by default (#12134) 2022-07-14 14:10:51 +04:00
Tingqian Li
b7b3f0ab4a move cpu_dump_check into CPU plugin's tools folder (#12123) 2022-07-13 13:38:17 +08:00
Paul Youngsoo Ahn
0621e8cf28 [GPU] Fix gather data type issue (#12089) (#12089) 2022-07-12 19:01:07 +09:00
Tomasz Dołbniak
9d6d84088f Virtual destructor for the base class (#12103) 2022-07-12 11:55:41 +02:00
Eddy Kim
a63dad6fdd updated to fuse activation in eltwise_vload8 (#12092) 2022-07-12 18:51:48 +09:00
Wang, Yang
bbc1c26750 setting tput as the default performance mode only for AUTO, excluding MULTI plugin. (#12090) 
Signed-off-by: ywang2 <yang4.wang@intel.com>

Co-authored-by: Chen Peter <peter.chen@intel.com>
Co-authored-by: Shen, Wanglei <wanglei.shen@intel.com>
 2022-07-10 15:16:59 +08:00
Eddy Kim
8d852b4aee fixed 'is_rotating_except_batch' to follow the IE order (#12050) 2022-07-08 15:36:17 +09:00
Tingqian Li
bc34fa0934 [CPU] Re-enable Selective build on oneDNN2.6 (#12074) 
* update submodule onednn26 selective build

* onednn code review

* merge onednn selective build

* fix bug in cc onednn26

Co-authored-by: zihan wu <zihan.wu@intel.com>
 2022-07-08 03:48:12 +00:00
guozhong wang
ab8c2f6fd8 change gpunum to 3 (#12073) 2022-07-07 18:15:27 +03:00
Andrew Kwangwoong Park
32937ab7ca Add Debug Config for maximum kernels per batch (#12068) 
Signed-off-by: Andrew Park <andrew.park@intel.com>
 2022-07-07 14:26:51 +03:00
guozhong wang
cd6c7da91c AUTO/MULTI supports ov::auto_batch_timeout (#12023) 
* add auto_batch_timeout for MULTI and AUTO

* fix clang-format for ie_core.cpp

* fix coredump

* simplify insert key to deviceConfig logic and parseDeviceNameIntoConfig() check "AUTO" and "AUTO:" only

* check config auto_batch_timeout

* add CleanUpInIECore()

* fix clang-format for ie_core.cpp
 2022-07-07 10:33:04 +00:00
Luwei Zhou
0224e6a067 Fix the deconv depwise post ops issue on AVX2 and AVX512 and enable deconv test (#11870) 
* Fix the deconv fused issue on AVX2 and AVX512 and enable deconv test

* Keep GroupDeconv BF16 test cases still disabled.

* Update to also excluding nightly

* Update onednn submodule.

* Update onednn submodule

* Update onednn submodule.

* Update the ONDENN submodule

* Update the ONEDNN commit.

* Update with merged onednn commit.
 2022-07-07 13:26:44 +08:00
River Li
b80f724414 Fix rnn cache missing issue (#12053) 2022-07-06 11:20:27 +00:00
Kelvin Choi
63ab516c85 [GPU] Delete previous inputs by numbered new name for batching (#12045) 2022-07-06 16:32:14 +09:00
yanlan song
e718e51a85 Bell/fix lifecycle coredump (#11934) 
* enable binder schedule

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

* add cases

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

* refine

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

* fix build failure

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

* fix coredump

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

* do not return hw requests directly, potential issues

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

* fix bug

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

typo

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

* optimize memory

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

* hold the hw plugin

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

* Revert "hold the hw plugin"

This reverts commit 5b537f5b6f.

* apply the fix

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

apply the fix

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

* hold the plugin library for destructing tensor

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

* solve the virtuual plugin Getblob life cycle issue

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

* remove log

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

* refine interface

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

* fix build failure

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

* fix for hetero plugin

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

* replace with vector

* enable life time tests for virtual plugins

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

rework cases due to vpux build issue

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

disable context test for now

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

Co-authored-by: Chen Peter <peter.chen@intel.com>
 2022-07-06 05:21:17 +00:00
opoluektov-lohika
7a50ce2491 Coverity: fix issue with uninitialized members (#11996) 2022-07-05 23:55:53 +00:00
Mateusz Bencer
43c0c964b8 Added FoldSubgraphEmptyInputs transformation (#11957) 2022-07-05 19:38:46 +02:00