platform/upstream/dldt.git
3 years ago[IE CLDNN] Enabled tensor pitches in topk kernel for 0 input (#1898)
Vladimir Paramuzov [Mon, 24 Aug 2020 10:03:39 +0000 (13:03 +0300)]
[IE CLDNN] Enabled tensor pitches in topk kernel for 0 input (#1898)

3 years ago[IE CLDNN] Forcing bfzyx format in MVN layer (#1903)
Lukasz Debski [Mon, 24 Aug 2020 10:03:21 +0000 (12:03 +0200)]
[IE CLDNN] Forcing bfzyx format in MVN layer (#1903)

3 years agoReduce number of ops needed to create InstanceNorm (#1896)
Mateusz Tabaka [Mon, 24 Aug 2020 09:26:15 +0000 (11:26 +0200)]
Reduce number of ops needed to create InstanceNorm (#1896)

* Reduce number of ops needed to create InstanceNorm

InstanceNorm in onnx importer creates the same subgraph for Mean twice - once for Variance and once for actual Mean.
This change makes InstanceNorm to use single Mean which is shared by numerator and Variance.

Also enables IE_CPU.onnx_model_instance_normalization test case

* Revert changes to .gitignore

3 years ago[IE CLDNN] FP16 support for some quantized kernels (#1855)
Vladimir Paramuzov [Mon, 24 Aug 2020 05:47:45 +0000 (08:47 +0300)]
[IE CLDNN] FP16 support for some quantized kernels (#1855)

3 years agoLinux clang compilation error fix (#1891)
Tomasz Dołbniak [Fri, 21 Aug 2020 14:22:54 +0000 (16:22 +0200)]
Linux clang compilation error fix (#1891)

3 years ago[ MO ] Change layout of Shape sub-graphs once (#1875)
Evgenya Stepyreva [Fri, 21 Aug 2020 13:27:02 +0000 (16:27 +0300)]
[ MO ] Change layout of Shape sub-graphs once (#1875)

* [ MO ] Change layout of Shape sub-graphs once

3 years agoAccept empty initializers from ONNX models (#1776)
Tomasz Dołbniak [Fri, 21 Aug 2020 12:15:27 +0000 (14:15 +0200)]
Accept empty initializers from ONNX models (#1776)

* Accept empty initializers from ONNX models

* Create dummy constants instead of null nodes

* Code formatting

* Test the empty initializers handling in onnx importer

3 years ago[GNA] Additional checks (#1018)
Denis Orlov [Fri, 21 Aug 2020 11:43:01 +0000 (14:43 +0300)]
[GNA] Additional checks (#1018)

3 years agoAdded detailed error messages for cases when network has dynamic shapes on the conver...
Ilya Churaev [Fri, 21 Aug 2020 10:13:44 +0000 (13:13 +0300)]
Added detailed error messages for cases when network has dynamic shapes on the conversion moment (#1890)

3 years agoAdd method to classification_result for returning results (#1460)
Harald Rotuna [Fri, 21 Aug 2020 08:59:12 +0000 (11:59 +0300)]
Add method to classification_result for returning results (#1460)

3 years agoFix addOutput (#1888)
Gleb Kazantaev [Fri, 21 Aug 2020 02:51:42 +0000 (05:51 +0300)]
Fix addOutput (#1888)

3 years agoAzure CI: Enable all MklDnnFunctionalTests (#1881)
Alexander Zhogov [Thu, 20 Aug 2020 17:53:01 +0000 (20:53 +0300)]
Azure CI: Enable all MklDnnFunctionalTests (#1881)

* Azure CI: Enable all MklDnnFunctionalTests

* Update azure-pipelines.yml

* Disable failed test on Mac: moke_MobileNet/ModelTransformationsTest.LPT/mobilenet_v2_tf_depthwise_batch1_inPluginDisabled_inTestDisabled_asymmetric*

3 years agoFixed error message for ONNX importer tests (#1880)
Ilya Churaev [Thu, 20 Aug 2020 15:46:34 +0000 (18:46 +0300)]
Fixed error message for ONNX importer tests (#1880)

3 years ago[JAVA] Changed blob costructor (using cArray) in samples (#1863)
Anna Likholat [Thu, 20 Aug 2020 15:45:14 +0000 (18:45 +0300)]
[JAVA] Changed blob costructor (using cArray) in samples (#1863)

3 years agoReuse make_executable_network in ImportNetwork (#1869)
Ilya Lavrenov [Thu, 20 Aug 2020 15:34:19 +0000 (18:34 +0300)]
Reuse make_executable_network in ImportNetwork (#1869)

* Reuse make_executable_network for ImportNetwork

* make_executable_network returns ExecutableNetwork cpp wrapper

3 years agoUpdating ADE version (#1877)
Anton Potapov [Thu, 20 Aug 2020 15:32:41 +0000 (18:32 +0300)]
Updating ADE version (#1877)

- Fix gcc 9 build warnings

3 years ago[IE CLDNN] Add 3d spatials support to conv & pool imad kernels and unblock any in...
Jedrzej Hajduczenia [Thu, 20 Aug 2020 11:54:54 +0000 (13:54 +0200)]
[IE CLDNN] Add 3d spatials support to conv & pool imad kernels and unblock any in/out feature sizes (#1693)

3 years agoONNX ScatterND MO support (#1813)
Evgeny Lazarev [Thu, 20 Aug 2020 10:57:22 +0000 (13:57 +0300)]
ONNX ScatterND MO support (#1813)

* Added extractor for ONNX operation ScatterNDUpdate

* Updated list of supported operations for ONNX

3 years ago[IE TESTS] Added single layer tests HSwish (#1861)
Roman Lyamin [Thu, 20 Aug 2020 10:32:16 +0000 (13:32 +0300)]
[IE TESTS] Added single layer tests HSwish (#1861)

3 years ago[IE][VPU]: Execution graph via ngraph for VPU plugin (#1572)
Anton Dudchenko [Thu, 20 Aug 2020 09:56:12 +0000 (12:56 +0300)]
[IE][VPU]: Execution graph via ngraph for VPU plugin (#1572)

* Execution graph via ngraph for VPU plugin
* Enable ExecGraphTests
* Assert throw for multi device in tests
* Assert throw for gna device in tests

3 years ago[ MO ] FloorDiv replacement (#1860)
Evgenya Stepyreva [Thu, 20 Aug 2020 09:33:44 +0000 (12:33 +0300)]
[ MO ] FloorDiv replacement (#1860)

* [ MO ] FloorDiv replacement

* BOM file

* Review comments

* FloorDiv in SupportedFrameworkList.md

3 years agoRemove Structural Attributes from Proposal op. (#1853)
Ewa Tusień [Thu, 20 Aug 2020 09:31:43 +0000 (11:31 +0200)]
Remove Structural Attributes from Proposal op. (#1853)

3 years agoDeprecate nGraph v0 ops and builders (#1856)
Ilya Churaev [Thu, 20 Aug 2020 09:27:14 +0000 (12:27 +0300)]
Deprecate nGraph v0 ops and builders (#1856)

* Deprecate nGraph v0 ops

* Fixed build

* Added deprecated defines to fix windows

3 years ago[ MO ] KSO ON by default (#1730)
Evgenya Stepyreva [Thu, 20 Aug 2020 09:06:45 +0000 (12:06 +0300)]
[ MO ] KSO ON by default (#1730)

* [ MO ] KSO ON by default.
Introduce static_shape MO key, remove keep_shape_ops key usages

* keep_shape_ops -> static_shape

* static_shape = False by default

* Fix Proposal-4 shape infer

* style-apply

* Fix Layout Change for ShapeOf paths

3 years agoRevert "[GNA] Remove old GNA1 from Cmake scripts (#1686)" (#1864)
Pavel Rodionov [Thu, 20 Aug 2020 07:59:21 +0000 (10:59 +0300)]
Revert "[GNA] Remove old GNA1 from Cmake scripts (#1686)" (#1864)

This reverts commit ffe8599c309b32ef4bf6b6e73ea7a2edd2c048c0.

3 years agoinstall_NEO_OCL_driver.sh: Added downloading driver functionality after removal NEO...
Artyom Anokhov [Wed, 19 Aug 2020 19:21:34 +0000 (22:21 +0300)]
install_NEO_OCL_driver.sh: Added downloading driver functionality after removal NEO from the package (#1872)

3 years agoRestore change of i64 parameters to i32 (#1866)
Evgeny Lazarev [Wed, 19 Aug 2020 17:18:04 +0000 (20:18 +0300)]
Restore change of i64 parameters to i32 (#1866)

3 years agoChange "run_op_node" helper to use Parameter instead of Constant (#1722)
Ewa Tusień [Wed, 19 Aug 2020 15:55:20 +0000 (17:55 +0200)]
Change "run_op_node" helper to use Parameter instead of Constant (#1722)

* Replace Constant with Parameter in run_op_node.

* Pass inputs to function.

* Add func to get shape.

* Make constant if input is scalar.

* Add case for list.

* Fix test.

* Split tests for run_op_node and run_op_numeric_data.

* Split more tests.

* Split more and more tests.

* Mark tests with xfail.

* Mark more tests with xfail.

* Replace scalar with parameter.

* Code formatting.

* Set empty shape for scalar.

* Remove check for list.

3 years agoMoved AlgebraicSimplification and NopElimination passes to IE (#1859)
Gleb Kazantaev [Wed, 19 Aug 2020 15:55:11 +0000 (18:55 +0300)]
Moved AlgebraicSimplification and NopElimination passes to IE (#1859)

* Moved AlgebraicSimplification and NopElimination passes to IE

* Fixed headerfiles

3 years ago[IE CLDNN] Enable fsv16 format for all quantized models (#1791)
Vladimir Paramuzov [Wed, 19 Aug 2020 13:19:42 +0000 (16:19 +0300)]
[IE CLDNN] Enable fsv16 format for all quantized models (#1791)

3 years ago[NGRAPH] Fix system protoc version check in CMake (#1852)
Vladislav Vinogradov [Wed, 19 Aug 2020 13:05:00 +0000 (16:05 +0300)]
[NGRAPH] Fix system protoc version check in CMake (#1852)

Add `OUTPUT_STRIP_TRAILING_WHITESPACE` option to `execute_process` command.

Latest CMake (tested 3.18.1) doesn't strip new line from `protoc --version` call,
which leads to wrong `PROTOC_VERSION` variable and failure on git fetch step.

3 years agore-enable the myriadMultipleGraphsTests_nightly (#1836)
Evgeny Latkin [Wed, 19 Aug 2020 13:03:37 +0000 (16:03 +0300)]
re-enable the myriadMultipleGraphsTests_nightly (#1836)

* re-enable the myriadMultipleGraphsTests_nightly

* refresh version of Myriad firmware

3 years ago[GNA] add support for 4d eltwise (#1353)
Anna Alberska [Wed, 19 Aug 2020 13:03:18 +0000 (15:03 +0200)]
[GNA] add support for 4d eltwise (#1353)

* [GNA] add support for 4d eltwise

* cpplint fix

* refactor

* fix FP16 broadcast pass

* change anonymous function to InferenceEngine::details::product

* introduce anonymous function to the pass & add layer checks

* refactor tileBlob lambda

3 years ago[IE CLDNN] Fix vector iterators incompatibility issue in prepare_primitive_fusing...
Jedrzej Hajduczenia [Wed, 19 Aug 2020 12:51:24 +0000 (14:51 +0200)]
[IE CLDNN] Fix vector iterators incompatibility issue in prepare_primitive_fusing (#1842)

3 years agoFixed addOutput (#1844)
Gleb Kazantaev [Wed, 19 Aug 2020 11:57:28 +0000 (14:57 +0300)]
Fixed addOutput (#1844)

* Fixed addOutput

* Added tests; removed addOutput second call

3 years ago[IE CLDNN] Mixed precision scale support (#1848)
Vladimir Paramuzov [Wed, 19 Aug 2020 11:17:45 +0000 (14:17 +0300)]
[IE CLDNN] Mixed precision scale support (#1848)

3 years agoRemoved MO conversion of i64 inputs to i32 (#1835)
Evgeny Lazarev [Wed, 19 Aug 2020 10:58:57 +0000 (13:58 +0300)]
Removed MO conversion of i64 inputs to i32 (#1835)

3 years agoExtractImagePatches MO implementation and nGraph transformation (#739)
Anton Chetverikov [Wed, 19 Aug 2020 10:23:41 +0000 (13:23 +0300)]
ExtractImagePatches MO implementation and nGraph transformation (#739)

3 years ago[ MO ] Random Uniform Replacer (#1814)
Yegor Kruglov [Wed, 19 Aug 2020 09:32:16 +0000 (12:32 +0300)]
[ MO ] Random Uniform Replacer (#1814)

3 years agoLoadNetwork interface: executable network is returned as return value (#1840)
Ilya Lavrenov [Wed, 19 Aug 2020 08:44:29 +0000 (11:44 +0300)]
LoadNetwork interface: executable network is returned as return value (#1840)

* LoadNetwork interface

* Added reference to SOPointer to returned ExecutablNetwork

* Fixed test

3 years ago[LPT] Symmetric quantization support for efficientnet-b0 (#1773)
Edward Shogulin [Wed, 19 Aug 2020 08:36:48 +0000 (11:36 +0300)]
[LPT] Symmetric quantization support for efficientnet-b0 (#1773)

* [LPT] Symmetric quantization for efficientnet-b0

* [LPT] zero point calculation fix

Co-authored-by: Zinoviev, Vladimir <vladimir.zinoviev@intel.com>
3 years ago[IE CLDNN] Use fsv4 instead of af32 layout for imad case (#1784)
Vladimir Paramuzov [Wed, 19 Aug 2020 07:10:55 +0000 (10:10 +0300)]
[IE CLDNN] Use fsv4 instead of af32 layout for imad case (#1784)

3 years agoFixed path for pad reference impl (#1846)
Ilya Churaev [Wed, 19 Aug 2020 06:41:59 +0000 (09:41 +0300)]
Fixed path for pad reference impl (#1846)

3 years ago[IE CLDNN] added fusing support for ref and b_fs_yx_fsv16 implementations (#1692)
Sergey Nesterov [Wed, 19 Aug 2020 06:18:00 +0000 (08:18 +0200)]
[IE CLDNN] added fusing support for ref and b_fs_yx_fsv16 implementations (#1692)

3 years agoEnable HSwish - ngraph op, fusion/decomposition and reference implementation (#1770)
Katarzyna Mitrus [Wed, 19 Aug 2020 05:04:00 +0000 (07:04 +0200)]
Enable HSwish - ngraph op, fusion/decomposition and reference implementation (#1770)

* Add HSwish operator to nGraph

* Add HSwishFusion transformation

* Update check_constant function

* Add reference implementation for HSwish

* Enable reference implemenation in HSwish evaluate

* Add op_eval test

* HSwish fusion transformation test

* Add HSwishFusionWithoutRelu transformation

* Add more hswish fusion tests

* Register HSwishFusion pass in common_optimizations

* Update HSwish reference implementation

* Add HSwishFusion with Relu and Multiply

* Add HSwishDecomposition transformation pass

* Add HSwishDecomposition test

* Add HSwish op to ngraph python API

* Update HSwish fusion transformations

* Remove HSwishFusion from common optimizations

* Update hswish python API

* Add bf16 to evaluate hswish

* Update hswish python API

* Move hswish reference implementation

* UnaryElementwiseArithmetic inheritance

* Enable HSwish callback for clDNN

* Register HSwishDecomposition pass in ConvertOpSet1ToLegacy

* Enable HSwishFusion pass in common optimizations

* Use NGRAPH_RTTI_DECLARATION

* Moved python hswish test to the test_ops_unary

3 years agoAdd Range-4 support (#1552)
Maxim Vafin [Wed, 19 Aug 2020 04:58:41 +0000 (07:58 +0300)]
Add Range-4 support (#1552)

* Add Range-4 support

* Add type infer and IR Reader extender

* Implement ngraph Range-4

* Implement evaluate for Range-4

* Remove not needed types

* Resolve review feedback

* Add RTTI defines

3 years agoRemove ngraph::Lambda class, replace TensorIterator body with ngraph::Function (...
Ivan Tikhonov [Wed, 19 Aug 2020 04:09:32 +0000 (07:09 +0300)]
Remove ngraph::Lambda class, replace TensorIterator body with ngraph::Function (#1830)

* remove Lambda class, replace TensorIterator body with ngraph::Function

* Fix passing parameters from parent graph to subgraph

Co-authored-by: mbencer <mateusz.bencer@intel.com>
3 years agoReturn serialization to IR v7 under conditional compilation (#1820)
Ilya Lavrenov [Wed, 19 Aug 2020 03:53:17 +0000 (06:53 +0300)]
Return serialization to IR v7 under conditional compilation (#1820)

3 years agoFixed set/get batch size logic for scalar inputs (#1837)
Ilya Churaev [Wed, 19 Aug 2020 03:37:04 +0000 (06:37 +0300)]
Fixed set/get batch size logic for scalar inputs (#1837)

3 years agoOptimize Concat operation (#1812)
Mateusz Bencer [Tue, 18 Aug 2020 15:28:57 +0000 (17:28 +0200)]
Optimize Concat operation (#1812)

* 1d case optimization

* code refactor

* concat optimization

* removed using template for concat

* unit tests to concat constant folding

* synchro with current master

3 years agoConvert Pad to GroupConvolution transformation (#1826)
Gleb Kazantaev [Tue, 18 Aug 2020 14:40:45 +0000 (17:40 +0300)]
Convert Pad to GroupConvolution transformation (#1826)

* Added ConvertPadToConvolution pass

* Removed NasNet transformation from MO

* Renamed Conv to GroupConv; Added tests

* Fixed typo; Added RTTI

3 years agoNgraph: hide itt.hpp to src (#1834)
Ilya Lavrenov [Tue, 18 Aug 2020 13:50:37 +0000 (16:50 +0300)]
Ngraph: hide itt.hpp to src (#1834)

* Use proper include for log.hpp

* Hide itt.hpp to src folder

* Removed CodeWriter

3 years ago[GNA] Implement ImportNetwork from stream (#1743)
Anna Alberska [Tue, 18 Aug 2020 13:41:46 +0000 (15:41 +0200)]
[GNA] Implement ImportNetwork from stream (#1743)

* add ImportNetwork from stream

* add test for importing network from stream

3 years agoMerge changes from 2020.4.0.1 (#1809)
Denis Orlov [Tue, 18 Aug 2020 13:11:10 +0000 (16:11 +0300)]
Merge changes from 2020.4.0.1 (#1809)

* [GNA] Propagate QoS timeout to the calling app (#1188)

* [GNA] Support timeout value set in Wait (#1499)

* [GNA] stored request id for completed sync infer request in order to get status later using wait() (#1458)

* stored request id for completed async infer request in order to get it's status later

* preserved status not started for multiple sequential calls to wait()

Co-authored-by: Denis Orlov <denis.orlov@intel.com>
* [GNA] Fix callbacks (#1607)

Co-authored-by: Eugene Smirnov <eugene.smirnov@intel.com>
3 years agoAdd arithmetic operators for bfloat16 (#1831)
Ilya Churaev [Tue, 18 Aug 2020 10:49:30 +0000 (13:49 +0300)]
Add arithmetic operators for bfloat16 (#1831)

3 years agoFix a specification for CTCLoss operation (#1786)
Roman Kazantsev [Tue, 18 Aug 2020 10:47:59 +0000 (13:47 +0300)]
Fix a specification for CTCLoss operation (#1786)

3 years agov1::Pad evaluator (#1771)
Tomasz Dołbniak [Tue, 18 Aug 2020 10:43:54 +0000 (12:43 +0200)]
v1::Pad evaluator (#1771)

* v1::Pad reference implementation

* ut fix: pad_negative_exterior_1d

* ut fix: pad_negative_exterior_1d_check_limits & pad_edge_1d

* Code formatting

* ut fix: pad_edge_1d_top_neg & pad_edge_1d_top_neg_bigger_than_tensor

* More Pad UT fixes

* Pad UT fixes: REFLECT mode

* Fix all Pad UTs

* Switch Pad evaluation in INT backend

* Non-template solution to v1::Pad::evaluate

* Always create v1::Pad with 4 inputs

* VS compilation error fix

* Python test fix

* Remove the v0::Pad constant folding pass

* Some extra checks in v1::Pad evaluator

* Code formatting

* Remove an obsolete CF test

3 years agoAdd ONNX Round op (#1605)
Mateusz Bencer [Tue, 18 Aug 2020 10:10:23 +0000 (12:10 +0200)]
Add ONNX Round op (#1605)

3 years ago[IE][VPU]: Reverts nGraph TI Unrolling for VPU (#1824)
Gladilov, Gleb [Tue, 18 Aug 2020 09:09:06 +0000 (12:09 +0300)]
[IE][VPU]: Reverts nGraph TI Unrolling for VPU (#1824)

3 years ago[IE][VPU]: Fixed MX firmware memory type patching (#1825)
Evgeny Latkin [Tue, 18 Aug 2020 09:05:18 +0000 (12:05 +0300)]
[IE][VPU]: Fixed MX firmware memory type patching (#1825)

Fixing Myriad X firmware loader after RTEMS 5 related changes merged into MDK
* Fixed MX firmware memory type patching
* Update firmware
* disable myriadMultipleGraphsTests_nightly

Co-authored-by: Chekhonin Andrey <andrey.chekhonin@intel.com>
3 years agoConvertPrecision for element::boolean (#1772)
Gleb Kazantaev [Tue, 18 Aug 2020 08:10:24 +0000 (11:10 +0300)]
ConvertPrecision for element::boolean (#1772)

* Added bool to u8 conversion

* Added opset1::ShapeOf handler

* Added ReduceLogicalAnd/Or support in ConvertPrecision pass

* Moved static map inside function; Updated callbacks

* Removed header

* Fixed tyle relaxed for cases when the same output consumes by multiple inputs in the same operation; added tests; fixed input types setting for already created type relaxed operations

3 years agoJava bindings (#506)
Anna Likholat [Tue, 18 Aug 2020 05:38:48 +0000 (08:38 +0300)]
Java bindings (#506)

* java bindings added

* bulid saples alongside with IE build

* fixes

* fixes

* fixed comments

* fixes

3 years agoRemoved NGRAPH_API from reference library (#1821)
Ilya Lavrenov [Mon, 17 Aug 2020 19:46:12 +0000 (22:46 +0300)]
Removed NGRAPH_API from reference library (#1821)

3 years ago[FIX] Fix data layout for reshaped network (#1748)
Tomasz Socha [Mon, 17 Aug 2020 19:14:39 +0000 (21:14 +0200)]
[FIX] Fix data layout for reshaped network (#1748)

* [FIX] Fix data layout for reshaped network

* [PATCH] Don't change compatible layouts

* Add UT for reshaped network

* FIX no. 1

3 years agoAdd static library with nGraph reference implementations (#1810)
Ilya Churaev [Mon, 17 Aug 2020 16:43:11 +0000 (19:43 +0300)]
Add static library with nGraph reference implementations (#1810)

* Removed reference implementations from public API

* Remove coordinate_transform from public API

* Introduced static library with reference implementations

3 years agoExtend MO for operation CTCLoss and partly refactor CTCGreedyDecoder (#588)
Roman Kazantsev [Mon, 17 Aug 2020 16:19:59 +0000 (19:19 +0300)]
Extend MO for operation CTCLoss and partly refactor CTCGreedyDecoder (#588)

* Extend MO for operation CTCLoss

* Change sequence length format to a mask format

* Add fixes after first-round review

* Add fixes after the second-round review

* Fixing CTCLossPlusCTCGreedyDecoder transformation

3 years ago[IE][VPU]: Refactor vpu configs (#1334)
Maksim Doronin [Mon, 17 Aug 2020 14:30:38 +0000 (17:30 +0300)]
[IE][VPU]: Refactor vpu configs (#1334)

* [IE][VPU]: Refactor vpu configs

* [IE][VPU][GT]: Refactor vpu configs usage in GT

* [IE][VPU]: Refactor vpu configs usage in myriad plugin

* [IE][VPU]: Refactor vpu configs usage in tests

* [IE][VPU]: Refactor vpu configs usage in deprecated_tests

* [IE][VPU]: Refactor vpu configs usage in tools

* [IE][VPU]: Use hddl metric vars instead of macros

* [IE][VPU]: style fixes

* [IE][VPU]: Replace MYRIAD_FORCE_RESET with MYRIAD_ENABLE_FORCE_RESET

* [IE][VPU]: Resolve confilcts with MYRIAD_TILING_CMX_LIMIT_KB

* [IE][VPU]: Remove MYRIAD_PLATFORM from tools

* [IE][VPU]: Fix DISABLE_REORDER config in StridedSlice

* [IE][VPU]: Redesign config headers hierarchy

* [IE][VPU]: Clarify comments related to vpu configs

* [IE][VPU]: Leave default values of deprecated configs

* [IE][VPU]: Revert beh tests on VPU_MYRIAD_PLATFORM

* [IE][VPU]: Revert beh tests with deprecated configs

* [IE][VPU]: Resolve conflicts

* [IE][VPU]: Revert opencv dependency

3 years agoEnable ReduceL1 and ReduceL2 operations (#1799)
Evgeny Lazarev [Mon, 17 Aug 2020 13:32:53 +0000 (16:32 +0300)]
Enable ReduceL1 and ReduceL2 operations (#1799)

* Initial version of ReduceL1, ReduceL2 and ReduceLp enabling in the MO

* Added operations ReduceL1 and ReduceL2 to nGraph

* Removed ReduceLp. Added ReduceL1 and ReduceL2

* Separated specification of ReduceLp into ReduceL1 and ReduceL2

* Updated ReduceL1 and ReduceL2 specification

* Fixed ReduceL1 and ReduceL2 type prop tests

* Implemented nGraph transformation to decompose ReduceL1 and ReduceL2. Disabled them for CPU and GPU plugins

* Updated supported framework layers

* Added unit tests for ReduceL1 and ReduceL2 reference implementation

* Fixed ReduceXXX operations reference implementation by adding support for a new parameter 'keep_dims'

* Fixed constant folding for v0::Any

* Added ReduceL1 and ReduceL2 to Python API

* Implemented ReduceL1 and ReduceL2 decomposition tests and fixed ReduceL2 decomposition

* Added specific creator for ReduceXXX operations instead of NodeBuilders

* Fixed conversion ReduceXXX to CNNLayer

* Fixed parser for ReduceLogicalXXX operations

3 years ago[IE CLDNN] Export runtime graph as nGraph function (#1684)
Mikhail Letavin [Mon, 17 Aug 2020 13:27:33 +0000 (16:27 +0300)]
[IE CLDNN] Export runtime graph as nGraph function (#1684)

3 years agoFGTT for CF (#1792)
Gleb Kazantaev [Mon, 17 Aug 2020 12:39:58 +0000 (15:39 +0300)]
FGTT for CF (#1792)

* Added FGTT for ConstantFolding

* Added test for CF disabling

3 years agoFix problem with renaming outputs from network (#1546)
Maxim Vafin [Mon, 17 Aug 2020 11:38:41 +0000 (14:38 +0300)]
Fix problem with renaming outputs from network (#1546)

3 years ago[IE CLDNN] All input layouts support in ref pooling (#1782)
Vladimir Paramuzov [Mon, 17 Aug 2020 11:34:49 +0000 (14:34 +0300)]
[IE CLDNN] All input layouts support in ref pooling (#1782)

3 years ago[CPU] fix conv + concat case and add tests (#1747)
Maxim Andronov [Mon, 17 Aug 2020 11:25:53 +0000 (14:25 +0300)]
[CPU] fix conv + concat case and add tests (#1747)

3 years agoRemoved template headers for Fused Ops (#1780)
Ilya Churaev [Mon, 17 Aug 2020 10:37:00 +0000 (13:37 +0300)]
Removed template headers for Fused Ops (#1780)

3 years ago[Python API] Deprecate IENetLayer class (#1731)
Anastasia Kuporosova [Mon, 17 Aug 2020 09:41:54 +0000 (12:41 +0300)]
[Python API] Deprecate IENetLayer class (#1731)

* [Python API] Deprecate IENetLayer class

* update samples

* undo changes in samples

3 years ago[IE CLDNN] Set arguments once (#1699)
Vladimir Paramuzov [Mon, 17 Aug 2020 09:39:30 +0000 (12:39 +0300)]
[IE CLDNN] Set arguments once (#1699)

3 years agoAdd reference implementation to VariadicSplit (#1698)
Mateusz Bencer [Mon, 17 Aug 2020 08:13:06 +0000 (10:13 +0200)]
Add reference implementation to VariadicSplit (#1698)

* implementation of variadic split ref impl

* Added tests

* added unit tests

* refactor

* review remarks

* fixed ie tests

3 years ago[IE][VPU][GT]: fix pool 3d to 2d reduction (#1760)
Evgeny Latkin [Mon, 17 Aug 2020 08:02:00 +0000 (11:02 +0300)]
[IE][VPU][GT]: fix pool 3d to 2d reduction (#1760)

* Fix minor error in pooling 3D to 2D conversion

3 years ago[IE CLDNN] Gemm floating point unit-tests fix (#1797)
Ilya Znamenskiy [Mon, 17 Aug 2020 07:09:18 +0000 (10:09 +0300)]
[IE CLDNN] Gemm floating point unit-tests fix (#1797)

3 years ago[IE CLDNN] Fix for DepthToSpace fusing into convolution (#1724)
Sergey Shlyapnikov [Mon, 17 Aug 2020 06:56:23 +0000 (09:56 +0300)]
[IE CLDNN] Fix for DepthToSpace fusing into convolution (#1724)

3 years ago[NGraph] CTCLoss reference implementation. (#1453)
Nikolay Shchegolev [Mon, 17 Aug 2020 03:10:51 +0000 (06:10 +0300)]
[NGraph] CTCLoss reference implementation. (#1453)

3 years agoRemove nGraph layout (#1763)
Ilya Churaev [Mon, 17 Aug 2020 03:05:08 +0000 (06:05 +0300)]
Remove nGraph layout (#1763)

3 years agoadd preliminary support of Proposal-4 in nGraph (#1448)
Pavel Esir [Sun, 16 Aug 2020 12:49:49 +0000 (15:49 +0300)]
add preliminary support of Proposal-4 in nGraph (#1448)

renamed logits -> bbox_deltas

updated ngraph unittests for Proposal

removed validate_and_infer_types Proposal-4

removed validate_and_infer_types Proposal-4

changed validate_and_infer_types in parent class of Proposal

removed get_output_size

successfully inferred Proposal on SSH and Faster-RCNN

added unittests for Proposal-4

added unittests for Proposal-4

added unittests for Proposal-4

returned back default namespace for Proposal

reduced number of outputs in v0::Proposal

correct conversion of Proposal-4 -> propodal_ie with 2 outputs

removed creator for proposal v0

removed converter for proposal v0

added Proposal-4 to MO

removed `for_deformable` attribute

added Proposal-4 to MO and nGraph Python API

removed typo in Proposal-4 specification

style corrections

style corrections and removed some redundant code

rename proposal Python api test

removed 'attrs' context from visitor

returned back AttrVisitor to check if passes OpenVINO ONNX pipeline

Should pass OpenVINO ONNX pipeline (returned back AttrVisitor just to check)

python api for Proposal-4 works ok

(style correction) python api for Proposal-4 works ok

parametrized proposal_ie some other corrections

removed 'attrs.' context from nGraph Python API tests for Proposal

minor corrections in replacer proposal->proposal_ie

corrected Python API OpenVINO-ONNX tests should pass

Improved workaround for AttributeVisitor for Proposal

Add additional check of im_info tensor shape to Proposal node in MKLDNNPlugin

😠 removed 4 extra spaces from test_dyn_attributes.py to match The Style

added new nGraph RTTI declarations, removed throwing exception in transformation

added new nGraph RTTI declarations, removed throwing exception in transformation, corrected exception in MKLDNNplugin

corrected im_info size checking in Proposal node of MKLDNNPlugin

3 years agoAdded split test into template plugin (#1669)
Anton Pankratv [Sat, 15 Aug 2020 11:44:13 +0000 (14:44 +0300)]
Added split test into template plugin (#1669)

* Added split test into template plugin

* Fixed split test

* Fixed test parameters style

3 years agoAzure CI: Pipeline improvements (#1790)
Alexander Zhogov [Fri, 14 Aug 2020 22:31:34 +0000 (01:31 +0300)]
Azure CI: Pipeline improvements (#1790)

* Azure CI: Pipeline improvements

* Fix checkout

* Fix path

* Fix "Install IncrediBuild"

* Add "submodules: recursive"

* Fix test run

* Fix work dir

* Fix issue

* Increase timeout

3 years agoAdded OpenVINO Get Started Guide for Windows (#1631)
Anastasiya Ageeva [Fri, 14 Aug 2020 15:47:58 +0000 (18:47 +0300)]
Added OpenVINO Get Started Guide for Windows (#1631)

3 years agoRemoved deprecated MulAddFusion pass (#1789)
Gleb Kazantaev [Fri, 14 Aug 2020 15:35:11 +0000 (18:35 +0300)]
Removed deprecated MulAddFusion pass (#1789)

3 years ago[IE][VPU][GT]: Add pass for marking fast stages (#1579)
Andrew Bakalin [Fri, 14 Aug 2020 15:10:57 +0000 (18:10 +0300)]
[IE][VPU][GT]: Add pass for marking fast stages (#1579)

It was discovered that it is sometimes useful to mark fast stages (e.g. stages that process less than 100 elements) in order to be able to parse them from final performance report and estimate its contribution into performance.
* Added pass for marking fast stages
* Introduced unit tests

3 years ago[IE CLDNN] Disable non-performant fp32 batches FC layer (#1737)
Alexander Chaiko [Fri, 14 Aug 2020 13:02:37 +0000 (15:02 +0200)]
[IE CLDNN] Disable non-performant fp32 batches FC layer (#1737)

3 years ago[IE CLDNN] dp4a check that works both with old and new drivers (#1766)
Mikhail Letavin [Fri, 14 Aug 2020 11:50:33 +0000 (14:50 +0300)]
[IE CLDNN] dp4a check that works both with old and new drivers (#1766)

3 years agoopset4 Convolution/GroupConvolution -> Multiply fusion (#1754)
Gleb Kazantaev [Fri, 14 Aug 2020 10:47:02 +0000 (13:47 +0300)]
opset4 Convolution/GroupConvolution -> Multiply fusion (#1754)

* Added new predicates for smart pattern matching

* Added ConvMul and GroupConvMul fusion passes based on opset4; Added CPU functional tests for comparing fusion accuracy

* Improved ConvMultiply fusion to support scalars; Added positive and negative tests

* Added ConvolutionBackprop/GrouConvolutionBackprop Multiply fusion; Added functional tests

3 years agoSimplified plugin interfaces (#1745)
Ilya Lavrenov [Fri, 14 Aug 2020 09:11:54 +0000 (12:11 +0300)]
Simplified plugin interfaces (#1745)

* Simplified plugin interface

* Allow not implemented

* Fixes

* Fixed CPU plugin tests

* Fixed tests dependencies

* Fixes

* Fixed GPU plugin compilation

* Renamed plugin

* Fixes

* Removed tests for plugin base

* Fix2

* Fix 2

* Define a macro to define plugin creation function

* Clean-up

* Fixed OSX build

* Fixed CentOS

* Fixed exception catch / throw

* Fixed clang issue

* Fixed python tests on macOsx

3 years agoAdd asinh acosh atanh to python api (#1488)
iliya mironov [Fri, 14 Aug 2020 07:07:58 +0000 (10:07 +0300)]
Add asinh acosh atanh to python api (#1488)

* Add asinh acosh atanh to python api

3 years agoDeprecate FusedOp class (#1758)
Ilya Churaev [Fri, 14 Aug 2020 03:27:58 +0000 (06:27 +0300)]
Deprecate FusedOp class (#1758)

* Deprecate FusedOps

* Try to fix windows

* Added temp headers

3 years agoAzure CI: Add IncrediBuild option: /MaxCPUS=40 (#1779)
Alexander Zhogov [Thu, 13 Aug 2020 21:55:34 +0000 (00:55 +0300)]
Azure CI: Add IncrediBuild option: /MaxCPUS=40 (#1779)

* Azure CI: Add IncrediBuild options: /ShowTime /StopOnErrors /MaxCPUS=62

* set 48 cores

* Remove /StopOnErrors

* Set 40 cores

3 years agoUpdated mock interfaces in tests (#1762)
Ilya Lavrenov [Thu, 13 Aug 2020 17:17:30 +0000 (20:17 +0300)]
Updated mock interfaces in tests (#1762)

* Updated mock interfaces in tests

* Added mock_engine dependency

3 years ago[nGraph] Add Manager to Py API (#1533)
Jan Iwaszkiewicz [Thu, 13 Aug 2020 16:56:59 +0000 (18:56 +0200)]
[nGraph] Add Manager to Py API (#1533)

* Added test

* working ManagerWrapper

* Clean-up in ManagerWrapper

* worksave

* fixed building error

* Finished test of constant folding

* remove unused param

* Added get_vector function

* clean up

3 years ago[CPU] Add explicit storage for MemoryNode (#895)
Alexander Peskov [Thu, 13 Aug 2020 16:06:20 +0000 (19:06 +0300)]
[CPU] Add explicit storage for MemoryNode (#895)

3 years agoTypeRelaxed implementation (#1561)
Sergey Lyalin [Thu, 13 Aug 2020 15:45:37 +0000 (18:45 +0300)]
TypeRelaxed implementation (#1561)

* RTTI base for ngraph::Node; cherry-pick from another branch, draft

* Added comments, moved code, switched to custom RTTI-based version of is_type

* Move rtti definitions in ngraph op class to the beginning of each class definition as a preparation for the next replacement

* Migrate part of operations to new RTTI

* Migrate GroupConvolution and Concat to new RTTI

* Apply code style for ngraph part

* Rename RTTI_DECLARATION/DEFINITION to NGRAPH_RTTI_DECLARATION/DEFINITION

* Reverted accidentally updated version of mkldnn

* TMP: rewrite RTTI back to constexprions as an attempt to fix static objects initialization order issue

* Apply ngraph code style

* Finalize move back to constexpr for RTTI

* Applied code-style

* TypeRelaxed template class implementation and necessary changes in ngraph + tests.

* Applied code-style

* Fix in fast algorithm in GraphRewrite, add new tests for this and other cases

* Make parent optional parameter for NGRAPH_RTTI_DECLARATION and remove Node::type_info; remove ability to have Node as a parent for type_info

* Try to resolve compilation error on Windows

* The next attempt to fix Windows build: re-introduce get_type_info_static

* Removed file that was removed in master and kept in this branch by mistake

* Next attempt to fix Windows build: externConstexpr

* Attempt to fix win build: extra public (suspect icc bug), remove get_type_info_static as useless.

* Next attempt to fix Windows: proxy const and constexpr

* Fixed constexpr

* Next attmpts: move get_type_info to cpp file

* Code stype fix

* Re-implemented RTTI without use of constexpr; run-time initialization is used; removed global definitions to avoid issues with order of static objects initialization

* Removed externConstexpr flag and removed TRANSFOMRATIONS_API for TypeRelaxed

* get_type_info_static initializes static local constant with type_info that is used for CLASS::type_info and CLASS::get_type_info

* Removed not needed debug output and useless comments

* Implemented better copy ctor for Node

* Fixed VisualizeTree issue for TypeRelaxed: stopped using < and > in type_info::name

* Better comments and names for methods

* Remove unused include

* Remove commented line

* Workaround for legacy conversion that uses Node::get_type_info().name as a type for the resulting CNNLayer leading to incorrect types for TypeRelaxed-based operations and then to fail in plugins

* Fixed typos, explicit ctor for TypeRelaxedBase, explanation for the need of get_overridden_output_type

* Fix typo

* Fixed issue with non-static name in type definition for TypeRelaxed and fixed WrapType to make it compatible with hierarchical relations between types

* Reverted default ctor for Output and reverted ability to reduce number of outputs for a Node; syntactically better debug message for a Node

* Cover methods of TypeRelaxedBase by tests

* Apply code-style

3 years ago[IE][VPU][GT]: Fix different blobs for the same network (#1738)
Andrew Bakalin [Thu, 13 Aug 2020 11:57:10 +0000 (14:57 +0300)]
[IE][VPU][GT]: Fix different blobs for the same network (#1738)

* Use vector instead of unordered_map in order to get stable blob serialization.