e8fa1aaccb
fixed crash related to loading model with fq and sigmoid ( #6644 )
...
* fixed crash related with loading model with fq and sigmoid
* renamed multiple_input.* to multiple_input_fq.*; removed two unnecessary FQ layers from smoke_fq_fusion_with_sigmoid test; moved FQ params to test params
2021-07-20 15:38:39 +03:00
ad852f78b4
[§] cleans snippets interface and adds subgraph tests ( #6123 )
2021-06-16 01:00:19 +03:00
64d7a40ae4
[IE CLDNN] 54304 fix reduce ops ( #5986 )
2021-06-07 18:36:38 +03:00
c1608628d4
LowLatency v2 ngraph transformation ( #5160 )
...
* LowLatency 2.0: transformation and unit tests
* low latency 2.0: unit tests
* documentation and ngraph codestyle
* update CNN Interface of LowLatency transformation
* fix build on Windows
* fix build on Windows
* investigation of a failed build on Win OS
* ngraph codestyle
* fix build (werrors)
* New unit tests, refactoring
* update functional tests for Memory
* update LowLatency functional tests
* extend Memory tests to cover LowLatency v2 transformation
* clean up, code style
* fix unit tests
* update and fix unit tests, add feature to apply LLTv2 after LLTv1
* update docs, refactoring
* add several gna tests to skip config
* fix python api tests
* update python api, rename LowLatency_v2 to LowLatency2
* deprecate LowLatency v1
* Deprecate LowLatency v1 in IE
* fix wrong merge, codestyle
* resolve review comments
* fix python test
* update skip config
* apply online review notes, fix unit tests
* clean up, code style
* fix docs
* Use debug_messages instead of exceptions in llt v2
* fix unit tests
* Resolve review remarks
2021-06-07 15:13:41 +03:00
487faf3f29
Fix 56171: Insert AffineFilter after split when padding is needed. Fix Basic_LSTM LLT test. ( #5882 )
2021-06-02 16:09:56 +03:00
4b486e6223
[GNA] Fix fq weights fusion and relu scale factors calculation ( #5886 )
...
* Fix FakeQuantize weights fusion when multiple operations were using the same const
* [GNA] Fix scale factor calculation for Relu if it has source and doesn't have destination statistics
Co-authored-by: Dmitrii Khurtin <dmitrii.khurtin@intel.com>
2021-06-02 10:11:11 +03:00
f450f61bc3
[GNA] Avoid integers overflow during pwl calculation for FakeQuantize ( #5841 )
...
* [GNA] Avoid integers overflow during pwl calculation for FakeQuantize
* The similar fix for Relu
2021-05-27 16:49:40 +03:00
b7c5edc944
Deprecated all ICNNNetwork methods ( #5488 )
...
* Deprecated ICNNNetwork
* ICNNNetwork deprecation
* Fixed comments
* More suppressions for ICNNNetwork
* Fixed C API compilation
* Deprecated ICNNNetwork methods only
* [IE CLDNN] Fixed cpplint for clDNN with Ninja generator
* Fixed compilation for ApiVersion with clang
Co-authored-by: Vladimir Paramuzov <vladimir.paramuzov@intel.com>
2021-05-25 07:11:55 +03:00
57d49f3215
[CPU] Added MVN fusion for case with constants inside ( #5644 )
2021-05-21 14:35:56 +03:00
606d0f363a
[GNA] Fix concat scale factors calculation ( #5454 )
...
* [GNA] Fix concat scale factors calculation
* Add tests with Input | Constant -> Strided Slices -> Concat
Topology:
Constant Parameter
| | | |
+---+ +---+ +---+ +---+
| | | |
SS_1c ... SS_Nc SS_1p ... SS_Np
| | | |
| +----+ +----+ |
| | | |
+-------------+ | | +-------------+
\ | | /
Concat
Legend:
SS == Strided Slice
* Apply review
Co-authored-by: Elizaveta Lobanova <elizaveta.lobanova@intel.com>
2021-05-17 11:59:34 +03:00
a19413c0c0
[CPU] Plugin migration on ngraph ( #4344 )
2021-05-06 19:49:24 +03:00
78cf72819b
[GNA] fix pwl for Sigmoid quantised with signed range ( #5271 )
2021-04-22 13:27:21 +03:00
69be1f1c02
[GNA] Prohibit 2D data reshaping for diagonal layer if it's the first functional layer in network ( #5097 )
2021-04-12 11:23:29 +03:00
27268f008e
[GNA] Fix transformation from NCHW to NHWC for 4d concat ( #4852 )
...
Add additional checks for transformation from NCHW to NHWC
2021-04-01 18:38:30 +03:00
a748c26fee
Align copyright notice in cpp and cmake source files (CVS-51320) ( #4950 )
2021-03-25 02:40:09 +03:00
704278e09e
[CPU][IE TESTS] Added VariadicSplit and Pad subgraph tests ( #4583 )
...
* [IE TESTS] Added VariadicSplit and Pad subgraph tests
* Added more connectedIndexes
2021-03-18 19:07:29 +03:00
a2b8b974b8
Simplified IE Exceptions Implementation ( #4258 )
2021-03-18 16:30:16 +03:00
39ae56d71f
[GNA] Move weights transposition from NCHW to NHWC to GNA pass after FQ passes ( #4614 )
2021-03-12 12:12:41 +03:00
fd311e60f5
[GNA] Map 2d convolution to 1d if kernel width/height is equal to input width/height ( #4471 )
...
Fix transposition info handling for a crop layer
Throw exception for 1d convolution padding
2021-03-11 15:22:33 +03:00
bf467e097f
[GNA] Fixed issue with multiple connections between pair of nodes (Split=>Concat) ( #3953 )
...
* fixed: gna split-concat-concat sequence issue
fixed: gna permute-conv-maxpool-permute issue
* fix issue with multiple connections between split|crop|memory & concat
fix issue with serializing IR V7 model
* fixed: gna split-concat-concat sequence issue
fixed: gna permute-conv-maxpool-permute issue
* fix issue with multiple connections between split|crop|memory & concat
fix issue with serializing IR V7 model
* fixed issues after rebase
* Fix for the test TEST_F(FP32NonQuantizedTest, LSTMCellPropagateForward)
Input x[96] = 0.1
Scaled input = 0.01
Affine output = 64 * 0.01 * 0.1 + 0.1 = 0.164
Sigmoid(Affine output) = 0.541
Tanh(Affine output) = 0.163
Sigmoid(Affine output)*Tanh(Affine output) = 0.088
Sigmoid(Affine output)*Scaled input = 0.005
H = Sigmoid(Affine output)*Tanh(Affine output) + Sigmoid(Affine output)*Scaled input = 0.093
tanh(H) = 0.093
Result = H + Sigmoid(Affine output)*Scaled input = 0.093 + 0.541*0.093 = 0.093 + 0.050 = 0.143
* Updated copyright date
* [GNA] Added tests for cases connection split->concat
* [GNA] Added fix memory -> concat case
* fixed inf loop during quantization
* fixed code formatting
* fixed failing test smoke_concat_quant_memory_requant/ConcatQuantDuringMemoryRequantTest.CompareWithRefs/netPRC=FP16_IS=128_HS=128_targetDevice=GNA
* fixed removed & mark from pass manager main for loop
* added split=>concat case to InsertCopyLayerPass
Co-authored-by: Andrey Dmitriev <andrey.dmitriev@intel.com>
2021-03-03 12:54:09 +03:00
96b7b9441e
[GNA] Fix issues with compilation and quantization of input-split-concat pattern ( #4489 )
2021-03-02 14:59:55 +03:00
6c522fe76b
Allow the framework to generate input data without infer request object ( #4117 )
2021-03-02 10:40:02 +03:00
b2b350b851
[GNA] Tests for clamp layer ( #4363 )
...
* [GNA] Tests for clamp layer
* [GNA] Fix comments
2021-03-01 11:26:02 +03:00
ca9540fda0
Removed getOVNameForOperation ( #4514 )
2021-03-01 07:13:34 +03:00
304c31d10e
[GNA] Enable Zero Padding ( #4252 )
...
* [GNA] Enable Zero Padding
Fix CNN2D stride order
Draft pooling
* Implement CNN2D FP32 mode in GNAPlugin
* Move padding flag to Policy
* Remove inactive output rotation code from CNN2D graph compiler routine (review)
* Add ConvReluSequence tests to gnaFuncTest as DISABLED
2021-02-24 19:03:40 +03:00
7d605e58e5
[GNA] Fix transposition to NHWC for models with splits and concats ( #4419 )
2021-02-20 18:16:11 +03:00
3f6dbb8a00
[GNA] 4D StridedSlice, Clamp, InsertIdentityLayerPass, SubstituteScaleShiftBroadCastPass fixes ( #4061 )
...
* fixed: StrideSlice/Crop was broken when crop axis != dim.size() - 1 & added support for tensor dim > 2
fixed: clamp had hard coded upper & lower boundary to -50/50
fixed: InsertIdentityLayerPass was adding not necessary identities that was leading to exception when 2 identities was added to convolution output
fixed: SubstituteScaleShiftBroadCastPass caused invalid input tensor dimension overwrite - instead of overwrite currently reshape is injected
* updated copyright header dates
* updated copyright date
* fixed braces not at the end of line
* fixed braces in if else sequence
* shortened line length
* fixed issues reported in review
* fixed test - model in test assumed clamped range <-5,5> however code had hard coded value <-50,50>, which test assumed as correct result
* fixed smoke_MemoryEltwiseReshapeConcatTest/MemoryEltwiseReshapeConcatTest
* [GNA] Added test for crop 4d
Co-authored-by: Andrey Dmitriev <andrey.dmitriev@intel.com>
2021-02-12 12:27:00 +03:00
61c72fe1d6
[VPU][Benchmark app] Support trivial networks ( #4060 )
...
* [BENCHMARK APP] Support I64 input filling
* [VPU] Support trivial networks
* [VPU] Add test cases
* [VPU][Samples] Review fixes
* [Benchmark] Support scalar generating in python benchmark app
* [VPU] Allocate shape for unused inputs robustly
* [VPU] Make processing trivial cases more efficient
* [VPU] Rename isUsed->isConsumed
* [VPU][Tests][Samples] Review fixes
* [VPU] Process trivial cases at the very beginning
* [VPU] Review fixes
2021-02-09 16:00:41 +03:00
4393525313
[GNA] 1D convolution support for native NCHW models ( #4067 )
2021-02-05 14:55:11 +03:00
748780e392
Supports tensor names with delimiter ( #4085 )
...
* Supports tensor names with delimiter
* Fixed comments
* Allow to use different delimiters
2021-02-04 06:38:05 +03:00
2ebae7cf30
Introduce the Broker API to map original framework names to OV ( #3800 )
...
* Added tests
* Fixed tests
* Added tests to check addOutput method
* Added support of port names in the IR
* Update copyrights
* Deprecate tensor name
* Fixed comments
* Enabled functional tests for GPU, GNA and Myriad
* Fixed get_tensor().get_names()
* Added unit test to check tensor names
* Fixed code style
* Skip add output test for GNA
* Added serialization support
* Added PythonAPI
* Fixed tests
* Fixed tests
* Fixed typo
* Try to disable GNA test
* Fixed tests
* Removed unused variables
* Fixed tests
* Update documentation
* Fixed comment
2021-01-29 06:28:31 +03:00
9cfe909e1e
Unused variables ( #3963 )
...
* Added -Wused-variable flag
* Fixes for clang compiler
* Removed wrong -Wno-error from protobuf compilation
* More fixes
2021-01-22 17:41:15 +03:00
36edd1fb38
[GNA] Fix handling of permutes around convolution for a case with 1 channel ( #3481 )
...
Fix of concat scale factors calculation causing infinite loop during memory layer requantization
Fix of copy layer unsupported batch size
2021-01-13 16:12:19 +03:00
241b0faea1
[IE CLDNN] NGraph integration into cldnn plugin ( #2506 )
...
Co-authored-by: Roman Lyamin <roman.lyamin@intel.com>
Co-authored-by: Mikhail Letavin <mikhail.letavin@intel.com>
2020-12-23 13:35:44 +03:00
602f8f2e08
[IE TESTS] Move SLT classes to `SharedTestClasses` lib & add serialization functionality to the common class ( #3431 )
...
* [IE TESTS] Changing functional test utils structure
* Example
* Remove extra
* Apply comments
* fixes
* [IE TESTS] Change the structure
* Continue
* step 3
* [IE TESTS] Complete transition single layer test classes
* [IE TESTS] Transition Subgraph
* Fix subgraph namespaces
* fix
* Apply comments
* latm fix
2020-12-15 22:32:00 +03:00
5da7e8dab8
Fixed tests compilation for Android ARM ( #3572 )
...
* Fixed tests compilation for Android ARM
* Added check for size_t
2020-12-15 11:51:17 +03:00
6888ffa328
Revise Reference Implementation of Range op ( #3409 )
...
* Range: Align operator with spec and add unit tests
* Range: Remove output shape from range ref impl signature
* Range: Exclude backend unit tests for CPU and GPU due to unsupported dynamic ops
* Range: Add single layer test class for Range-4
* Range: Add unit test for shape inference
* Range: Add unit tests for i32 and f32
* Range: Refactor Range v0 backend test and added test for f32 type
* Range: Add floating point tolerance in unit tests to avoid failures due to precision
* Range: Add subgraph tests for Range add element-wise
* Range: Refactor Range class for single layer tests and add range add element-wise test with truncated inputs
2020-12-08 16:05:00 +03:00
86e5461d4b
[GNA] Flatten trivial concatenations ( #3396 )
...
* [GNA] Flatten trivial concatenations
* [GNA] move test to subgraph tests
* Style
* style
* review
* fix
* fixes
2020-12-01 18:33:36 +03:00
6534d315a5
[IE TESTS] Remove extra dependencies from NgraphFunction lib ( #3222 )
...
* [IE TESTS] Remove extra dependencies from NgraphFunction lib
* [TEMPORARY] Create LPT ngraph lib
* refactoring ngraph_helpers
* Fix LTO
2020-11-27 16:48:39 +03:00
a889acec53
[GNA] Fixes for concat with multiple const inputs ( #3130 )
2020-11-27 15:52:44 +03:00
860fae2f27
[GNA] fix InsertCopyLayerPass ( #3058 )
...
* [GNA] fix InsertCopyLayerPass
* style
* [GNA] fixed Concat -> Memory case for InserCopyLayer
* Style
* [GNA] fix some FWL size mismatch cases
* [GNA] fix crop|split -> memory cases
* [GNA] One layer -> multiple concats fix tests
* style
* [GNA] move copy insertion after trivial non-functional layers
2020-11-24 20:08:09 +03:00
9b85d67cce
Add test for RESULT_NOT_READY returned from Wait() in async mode ( #2793 )
...
* Add shared test for RESULT_NOT_READY return from Wait() in async mode
* Instantiate test for RESULT_NOT_READY for GNA Plugin only
* Fix compile error
* Increase model size for the RESULT_NOT_READY test
* Reuse most of the test
* Apply review
- Fix typo
* Make the test deterministic
* Use callback timestamp
* Apply review
* Use promise and future
2020-11-23 16:18:17 +03:00
f86065ce7a
[GNA] Fixed incorrect output layout for convolution ( #2426 )
...
* [GNA] add permute for NCHW Convolution H=1
* [GNA] fix permute after convolution case
* [GNA] fix Reshape1DOps transformation case
* [GNA] fix rm_cnn4a tests
* [GNA] fix wsj model tests
* quick fix
* [GNA] fix sw_exact mode
* [GNA] Add rotateOutput for convolution op instead of permute
* [GNA] fix CI
* [GNA] apply changes from review
2020-11-23 12:23:57 +03:00
fc1a3ce2f1
[GNA] Fake quantization layer support for int-8 mode for GNA plugin ( #2937 )
...
* [GNA] added support for per-channel FakeQuantise layer
* [GNA] added quantisation types detection in FQ enabled networks, and added input scale factors detection from FQ connected to input layer
* added FakeQuantize callback that will be use to cast integer values stored as float in FakeQuantized layer
* fixed per-channel multiplier calculation for int8 case
* precision improvements for int8 fake quantization and support for propagating scale factors to activation layers
* added initial int16 support
* added support for fake quantize layer with many connected output layers and support for FQ data encoded as FP16
* added support for already quantized weights
* Shared single layer test
* Added subgraph test
* Fix comment
* int8
* Enabling FQ tests on GNA
Co-authored-by: Eugene Smirnov <eugene.smirnov@intel.com>
Co-authored-by: Andrey Dmitriev <andrey.dmitriev@intel.com>
2020-11-20 16:40:19 +03:00
a10f71feeb
[GNA] Added tests for negative memory layer offset and split - trivial permute - concat scenario ( #2989 )
2020-11-12 17:07:38 +03:00
bd091500cd
[GNA] Added Multiplication 1x1x1 ( #2470 )
...
* [GNA] Broadcast power primitive
[GNA] Broadcast power primitive
[GNA] Added Multiplication 1x1x1
[GNA] Broadcast power primitive
Added secondary changes
* Added fix test
2020-11-10 20:56:21 +03:00
4a0170dcb0
[GNA] fixed case when input connected to concat from left and const from right ( #2874 )
...
* Added test
* Added fix
* Fix comment
* Fix conflict
2020-11-10 16:37:03 +03:00
1220fb2fe8
[GNA] Add support for SubstituteSoftSign for pattern with addition ( #2789 )
...
* [GNA] add changes for SubstituteSoftSign for IRv10
* cpplint fix
* [GNA] add requested changes & change SoftSign test validation
* [GNA] SubstituteSoftsign refactor
* [GNA] enable 3d input tests for softsign
2020-11-09 16:40:31 +03:00
1c3208ffe0
Low Latency transformation ( #2869 )
...
* initial draft of adding sinks to ngraph::Function
* style fixes
* code style fixes
* code style fixes
* code style fix
* review fix+build fix
* code style fix
* fix build
* API changed according to latest discussion
* review fixes
* review fixes + tests
* initial draft of adding sinks to ngraph::Function
* style fixes
* code style fixes
* code style fixes
* code style fix
* review fix+build fix
* code style fix
* fix build
* API changed according to latest discussion
* review fixes
* review fixes + tests
* added 1 more ctor
* style fixes
* used new api in ir parser
* fixed build
* update low latency transformation, fix unroll transformation, add unit tests, modify subgraph tests
* fix low latency transformation
* Update low latency transformation, unit and sub-graph tests
* update LowLatency transformation and tests
* ngraph codestyle
* fix build, update description
* resolve review remarks
Co-authored-by: Svetlana Dolinina <svetlana.a.dolinina@intel.com>
2020-11-06 14:11:11 +03:00
3f4d8b49ff
Fix cases when identities inserted identites are separated by nonFunctional layers ( #2771 )
2020-11-03 10:49:58 +03:00
Dmitrii Khurtin