* change gpunum to 3
* hold threads for GPU for MULTI:GPU,CPU
* need to first check if there is a CPU in the device list
* use getNumberOfCPUCores to get CPU cores
* load GPU first
* assign the correct value to multiSContext->_devicePriorities
* load GPU first and load CPU last and set numa for CPU
* MULTI set CPU affinity to “NUMA” during load network
* Load the CPU last while maintaining the original device priority
* not using vector for CPU
* There is no user setting affinity in MULTI, and NUMA is set for the CPU
* pass key ENABLE_HYPER_THREAD to CPU plugin and merge xiaoxia PR
* set ENABLE_HYPER_THREAD to NO
* modify log
* Modify the code according to xiaoxia and wanglei comments
* Modify the code according to bell comments
* [TF FE] Break the cycle for NextIteration
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* Fix code-style
* Better to cut by outputs of operation creating the cycle
* Remove extra thing
* Add test model
* Add test for conversion of TF1 While
* Apply code-review feedback: use pointers and correct error message
* Remove extra check
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* fix custom setting nstreams doesn't work due to the logic changed in ApplyPerformanceHints
* add same limitation for latency mode
* change bind core method to binding full physical core first, then logical core
* update custom setting streams and threads
* fix corner case when threads_per_stream is 1
* fix with comments
* add condition of big core stream is zero when binding core
Co-authored-by: Shen, Wanglei <wanglei.shen@intel.com>
Co-authored-by: Chen Peter <peter.chen@intel.com>
It might not be available in some cross-toolchains.
Extract the code for switching to gold linker into separate
CMake utility function to avoid duplication between multiple CMake scripts.
* Remove test for unqiue op
Signed-off-by: andrei.kochin <andrei.kochin@intel.com>
* Remove test empty line
Signed-off-by: andrei.kochin <andrei.kochin@intel.com>
* Moved MO meta data to RT info.
* Undo wrong change.
* Reformat code.
* Code reformat.
* Corrected error message.
* Test correction.
* Test fix.
* Updated MO IR Reader.
* Added testing of meta data for MO IR reader.
* Code corrections.
* Default params removed from meta data.
* Test corrected, fixed paths cropping.
* Small fix.
* Renamed legacy_path to legacy_frontend.
* Fixed error.
* Added PartialShape from str constructor, added to_string() method.
* Small corrections.
* Moved shape parsing to PartialShape and Dimension classes.
* Fixed errors.
* Added tests for shapes with comma separator.
* Fixed tests.
* Test fixed.
* Fixed test.
* Fixed tests.
* Small fix.
* Removed usage of depricated method.
* Fix in MO IR reader.
* Tests fixed.
* Fix in MO IR reader.
* Test fixed.
* Shape class alignment.
* Code fixes.
* Code fixes.
* Used unnamed namespace.
* Undo change of Shape attribute adapter.
* Undo change of Shape attribute adapter.
* Removed unnecessary includes.
* Removed unnecessary code.
* Test fixed.
* Fix in MO IR reader.
* Moved check_all_digits() to ov::util.
* Moved check_all_digits() to src/core/src/.
* Clang format.
* Apply suggestions from code review
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
* Removed dimension_util file.
* Removed wrong change.
* Removed extra spaces from string representation of Shape.
* Test fixed.
* Added support of comma separator for value. Updated help.
* Returned old format shape serializing.
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
* Basic Unique impl
* Shape inference tests
* More type prop tests
* Dynamic input tests
* Visitor tests
* Code formatting
* Minor corrections
* Modern c++ trick to make MSVC happy
* Refactor
* Code review
* More tests with dynamic shapes
* Op check test for Unique
* Well, a whitespace
* Different way of out shape calculation
* Spec
* Axis values constraint in the docs
* Zero-dim handling
* opset10_dump test adjustment
* 0D implementation and some test infra adjustments
* More coverage for scalar inputs
* type dependent test case generators
* A working version for 1D with all outputs
* Tensor shape calculation helper
* 1D handling with sorted mode included
* Handling of no-axis cases
* Some tests for no-axis case
* Unique's evaluate method
* A working fallback for the CPU plugin
* evaluate() removal
* Corrected shape inference
* More type prop fixes
* Unified axis extraction helpers
* Cleanup and preparation for slices processing
* Slice descriptor struct
* Static shapes generation with axis
* N-D solution with axis handling
* Slices comparison impl
* Fix Windows compilation errors
* Apply defaulting principles
Co-authored-by: Tomasz Jankowski <tomasz1.jankowski@intel.com>
* [GPU] Move device memory allocation from preprocessing stage to runtime to fix incorrect behaviour in case of tensor swapping
* [GPU] Apply review comments
* Replace source of dynamic layout cheking
* [C API][DOC] correct the doxy group for ppp
Signed-off-by: xuejun <Xuejun.Zhai@intel.com>
* [C API][DOC] improve the C API doc with new PPP interface
Signed-off-by: xuejun <Xuejun.Zhai@intel.com>
Signed-off-by: xuejun <Xuejun.Zhai@intel.com>
* [AUTO] update property setting logic.
* Update core::set_property() logic to only support primary property for AUTO and MULTI.
* Separate AUTO and MULTI supported properties.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Add test case.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* 1. revert the changes for ie core.
2. Enable AUTO/MULTI only accepting its own properties.
3. Enable AUTO/MULTI accepting device properties passed from loadnetwork().
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Revert "[AUTO]Update the incorrect config test for Myriad (#13271)"
This reverts commit 0552d98802.
* MULTI only accepts its own properties that is same as AUTO currently.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Add test cases for AUTO/MULTI property test.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update to enable MULTI supporting hw primary property setting throw the compile_model().
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Remove the added test case for setting secondary property through set_property().
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* 1. For AUTO/MULTI, property setting will be passed via core::compile_model() instead of core::set_property().
2. update the logic to infer precision setting that will transform into secondary property setting to each hw device.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* set default value for nstreams when -d AUTO/MULTI and no nstreams setting from command line.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update code format.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Enable BA to support if -d AUTO:MULTI,xxx/MULTI:AUTO,xxx. while AUTO Plugin need to update the logic of generating supported config list to virtual device.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* move device configuration handling outside form function main.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Revert "move device configuration handling outside form function main."
This reverts commit ef77bfc602.
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Setting property performance hint to 'THROUGHPUT' and passing to executable network if no setting for of AUTO/MULTI.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
* Update.
Signed-off-by: Wang, Yang <yang4.wang@intel.com>
DisableConvertConstantFoldingOnConstPath was renamed to MarkDequantizationSubgraph
and modified such that it also marks Subtract and Multiply from dequantization subgraph
with 'DequantizationNode' attribute.
Ticket: 65489
Co-authored-by: Michal Lukaszewski <michal.lukaszewski@intel.com>
