* Removed back phase transformations related to IRv7
* Fixed setting value for the input port using the 'set_value' method
* Removed front and middle phase transformations related to IRv7
* Cleanup the rest of the Model Optimizer transformations from IRv7 specific transformations
* Final cleanup of the deprecated IR v7 related code
* Removed 'blobs_as_input' usage in the Model Optimizer.
* Removed function '_fuse_add' from the Model Optimizer since it is not used anymore.
* Removed 'keep_in_IR' node attribute for FakeQuantize ops in the MO
* Disabled failing gpu_engine.user_context test
* Fix build issue
Why:
* Enable to build OpenVINO.
This change addresses the need by:
* Adding include directories,
* Removing IE::inference_engine_c_api dependency.
* Remove IE::inference_engine_nn_builder reference.
Why:
* Enable to build OpenVINO.
This change addresses the need by:
* Removing IE::inference_engine_nn_builder dependency.
Co-authored-by: Ilya Lavrenov <ilya.lavrenov@intel.com>
For b_fs_yx_fsv16 format in reference kernel features for dispatch are
rounded to multiple of 16. This change adds correct check in kernel to
return work-items that are inside this dispatch padding.
Previously those work-items could corrupt memory expected to be filled
with 0s, and for parametrized activation due to bounds checking with
modulo operator they could have been corrupting actual layer output.
Issue: CVS-27672
* [IE VPU] Add evaluate method to DSR
* [IE VPU] Enable DSR_Reshape tests
* [IE VPU] Improvements in DSR op
* [IE VPU] Fix typo in copyBlobAccordingUpperBound
* [IE VPU] Support dynamic inputs
* [IE VPU] Use dynamic inputs in tests
* [IE VPU] Improve conditions in propogateDynamism pass
* [IE VPU] Fix Myriad2 tests via dosabling reorder
* [IE VPU] make error message more explicit
* [IE VPU] Fix Win compilation: std::stoi in <string>
* [IE VPU] Improve data transferring to work with ND tensors
* [IE VPU] Avoid ODR in myriad common test utils
* [IE VPU] Split code in propagate dynamism into separate methods
* [IE VPU] Simplify conditions in DSR parsing
* [IE VPU] Emplace data in initialStages when remove stage order
- Named structures in bmp.h to avoid MSFT compiler error
- Fix for non-void function with missing return statement to avoid Intel compiler error
- Enabled "smoke_ExportUsingFileNameImportFromStreamNoThrowWithDeviceName" test
- Fix for MvncTest
* Execution graph via ngraph for CPU plugin
* Fixes
* Migrated to VariantImpl instead of Parameter
* Reverted to dedicated ExecutionNode once again
* Re-use new execution graph in tests
* Fixed one more tests to use execution graph via ngraph::Function
* [IE][VPU]: Enables dynamic output from middle of network support
This feature is very useful for debugging dynamic networks.
Changes include modification of existing addCopyForOutputsInsideNetwork
pass to respect dynamic outputs and moving propagateDynamismToOutputs
pass after addCopyForOutputsInsideNetwork. The motivation for last change
is to avoid unnecessary copy stages due to not synchronized logic, because
previously:
* First in Front-End (parseDSR) we mark shape data object as output
* Then in propagateDynamismToOutputs we insert copy stage for that case.
It's necessary if shape data object had other consumers
* Then in convertShapeNotation we insert Gather consumer for output data object
* Finally, addCopyForOutputsInsideNetwork inserts one more copy stage to leave
output data object without consumers.
Signed-off-by: Gladilov, Gleb <gleb.gladilov@intel.com>
* [IE][VPU]: Replaces attrs.has + attrs.get with attrs.getOrDefault
* [IE][VPU]: Fixes setting IE-notation and converted-notation to the same data object
* Modifications to support fp16 networks in KMB-plugin
* StridedSliceIE is removed
* One function convertFunctionToICNNNetwork with default parameter
* Some little changes in function convertFunctionToICNNNetwork()
* Delete some spaces in code (style changes)
* Edit code style
* Edit code style one more
* Edit code style again
* Remove row with Transpose()
This change:
- extends concat in-place optimization for resample on input
- adds resample primitive int8 support for bilinear mode
- fixes some potential issues with offset calculations with in8
* [IE VPU] Enable variable number of inputs for ExpPriorGridGenerator layer
* [IE VPU] Add test cases for ExpPriorGridGenerator layer with less than three inputs
