This repository was archived by the owner on Jul 1, 2024. It is now read-only.
memory leak fixes #110
Merged
memory leak fixes #110
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
saipj
commented
Jun 24, 2021
saipj
commented
- Replacing TF APIs with numpy calls
- Adding disable_ovtf option for comparison
- Adding no_show option for video samples
- Code formatting
sspintel
force-pushed
the
sai/memory_leak_fix
branch
from
bf552fd to
308a31f
Compare
sspintel
approved these changes
Jun 29, 2021
Collaborator
sspintel
left a comment
sspintel
left a comment
There was a problem hiding this comment.
Changes look good. Tested for memleaks.
ck-intel
added a commit
that referenced
this pull request
Jul 12, 2021
* Added support for OneDNN FusedBatchNormEx and FusedConV2D (#112) * Added support for OneDNN FusedBatchNormEx and FusedConV2D * Code formating Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> * FusedDepthwiseConv2d and improved FusedConv2D support (#114) * Upgradation of TensorFlow version to v2.5.0 (#111) * patch updated for TFv2.5.0 * added math_op to skip list for gpu * TFv2.5.0 setup for builds and defaults * changed tf build location in ci * change abi1 tf links, add docker tf pip whls build patch Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Lazy Fallback (Support for Dynamic Fallback) (#113) * Initial implementation of dynamic fallback * Error check fix for dynamic fallback * Improved coverage for dynamic fallback * Code formating for dynamic fallback updates * dyn fb bugfix & dyn fb disabled by default * Fixed python api typo * Code cleanup in executable.cc * Code formating for dynamic fallback updates * tf 1.x support for dyn. fallback * Code formatting * Fix for freezing inf. with dyn. fb. when interop is 1 * memory leak fixes (#110) * memory leak fixes and adding more user options * fix flag options and code format * OCM submodule updated for fused ops (#115) * Updated code for some checks (#104) * checkmarx code fixes * checkmarx fix * checkmarx code fix * code format fix * Update build_utils.py * code format fix * add suggested checks (#119) * checks in py files * added exception catching for assert statement * fixed uninitialized variable issue (#120) * Aditya/doc update (#123) * installing opecv during venv setup * typo fix and removed opencv installation manually * Adding azure pipelines using docker (#122) * Removed shell as True with subprocess pipe (#127) * removed shell=True with subprocess pipe * applied code format * Initial TROUBLESHOOTING.md and updates on ARCHITECTURE.md (#126) * GUIDE document added, ARCHITECTURE updated for dyn. fb. * Filename change: GUIDE.md to TROUBLESHOOTING.md * AWS, Azure and Colab classification notebook added (#125) * AWS, Azure and Colab classification notebook added * colab notebook added for object detection * fixing path traversal issues (#129) Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com> Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> Co-authored-by: mohdansx <mohdx.ansari@intel.com> Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> Co-authored-by: Sai Jayanthi <sai.jayanthi@intel.com> Co-authored-by: Chandrakant Khandelwal <chandrakant.khandelwal@intel.com> Co-authored-by: adisan3 <adityax.sanivarapu@intel.com> Co-authored-by: bhadur <bhadur.a.sm@intel.com> Co-authored-by: Ambarish Das <ambarish.das@intel.com>
sspintel
added a commit
that referenced
this pull request
Jul 30, 2021
* Added support for OneDNN FusedBatchNormEx and FusedConV2D (#112) * Added support for OneDNN FusedBatchNormEx and FusedConV2D * Code formating Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> * FusedDepthwiseConv2d and improved FusedConv2D support (#114) * Upgradation of TensorFlow version to v2.5.0 (#111) * patch updated for TFv2.5.0 * added math_op to skip list for gpu * TFv2.5.0 setup for builds and defaults * changed tf build location in ci * change abi1 tf links, add docker tf pip whls build patch Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Lazy Fallback (Support for Dynamic Fallback) (#113) * Initial implementation of dynamic fallback * Error check fix for dynamic fallback * Improved coverage for dynamic fallback * Code formating for dynamic fallback updates * dyn fb bugfix & dyn fb disabled by default * Fixed python api typo * Code cleanup in executable.cc * Code formating for dynamic fallback updates * tf 1.x support for dyn. fallback * Code formatting * Fix for freezing inf. with dyn. fb. when interop is 1 * memory leak fixes (#110) * memory leak fixes and adding more user options * fix flag options and code format * OCM submodule updated for fused ops (#115) * Updated code for some checks (#104) * checkmarx code fixes * checkmarx fix * checkmarx code fix * code format fix * Update build_utils.py * code format fix * add suggested checks (#119) * checks in py files * added exception catching for assert statement * fixed uninitialized variable issue (#120) * Aditya/doc update (#123) * installing opecv during venv setup * typo fix and removed opencv installation manually * Adding azure pipelines using docker (#122) * Removed shell as True with subprocess pipe (#127) * removed shell=True with subprocess pipe * applied code format * Initial TROUBLESHOOTING.md and updates on ARCHITECTURE.md (#126) * GUIDE document added, ARCHITECTURE updated for dyn. fb. * Filename change: GUIDE.md to TROUBLESHOOTING.md * AWS, Azure and Colab classification notebook added (#125) * AWS, Azure and Colab classification notebook added * colab notebook added for object detection * fixing path traversal issues (#129) Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com> Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> Co-authored-by: mohdansx <mohdx.ansari@intel.com> Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> Co-authored-by: Sai Jayanthi <sai.jayanthi@intel.com> Co-authored-by: Chandrakant Khandelwal <chandrakant.khandelwal@intel.com> Co-authored-by: adisan3 <adityax.sanivarapu@intel.com> Co-authored-by: bhadur <bhadur.a.sm@intel.com> Co-authored-by: Ambarish Das <ambarish.das@intel.com>
rrajore
added a commit
that referenced
this pull request
Sep 27, 2021
* Added support for OneDNN FusedBatchNormEx and FusedConV2D (#112) * Added support for OneDNN FusedBatchNormEx and FusedConV2D Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> * FusedDepthwiseConv2d and improved FusedConv2D support (#114) * Upgradation of TensorFlow version to v2.5.0 (#111) * added math_op to skip list for gpu * changed tf build location in ci * change abi1 tf links, add docker tf pip whls build patch Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Lazy Fallback (Support for Dynamic Fallback) (#113) * Initial implementation of dynamic fallback * Error check fix for dynamic fallback * Improved coverage for dynamic fallback * dyn fb bugfix & dyn fb disabled by default * Fixed python api typo * Code cleanup in executable.cc * tf 1.x support for dyn. fallback * Fix for freezing inf. with dyn. fb. when interop is 1 * memory leak fixes (#110) * OCM submodule updated for fused ops (#115) * Updated code for some checks (#104) * Update build_utils.py * add suggested checks (#119) * checks in py files * added exception catching for assert statement * fixed uninitialized variable issue (#120) * doc update (#123) * installing opencv during venv setup * typo fix and removed opencv installation manually * Adding azure pipelines using docker (#122) * Removed shell as True with subprocess pipe (#127) * Initial TROUBLESHOOTING.md and updates on ARCHITECTURE.md (#126) * GUIDE document added, ARCHITECTURE updated for dyn. fb. * Filename change: GUIDE.md to TROUBLESHOOTING.md * AWS, Azure and Colab classification notebook added (#125) * collab notebook added for object detection * fixing path traversal issues (#129) * Code changes for models support (#130) * Added following op translations and changes: * NonmaxsuppresionV3, FusedConv2d with BiasAdd LeakyRelu, FusedConv2d with BiasAdd Add Relu * get_shape replaced with get_partial_shape where only rank is needed * deassign cluster modifed for dynamic ops * translation for fusedops with ELU added * dynamic input condition handled for Fill op * zero dim check updated * Updated TopkV2 translation for zero dim inputs * fix for execution errors and collab badge support added * fix build error during tensorflow installation (#132) * Fix for build error during tensorflow installation Properly check cmake and gcc versions Force re-install packages for them to take effect after every re-build * OV2021.4 build support enabled (#131) * Updated the code for OV2021.4 * OCM updated for OV2021_4 * Enabled OpenVINO 2021.4, but kept default build version as 2021.3 due to SetBlob issue Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Fixed bug in FusedBatchNormEx translation (#134) * Fix cmake and gcc version checks (#136) * Fix cmake and gcc version checks * review comments incorporated in colab notebooks * Added Op translations and tests (#137) * Added translation for Resize Nearest Neighbor, GatherNd and Round OP translations * Added CropAndResize op and flip crop fix for CropAndResize * Added Reverse, Reciprocal, BatchToSpaceND, SpaceToBatchND, and Elu * Added TF Python tests for the above * Handle corner case in TranslateBatchNDAndSpaceNDOp * Add ops to gpu and myriad devices; update ocm * Adjust tolerance precision for fusedMatMul * Properly handle dynfallback flag; Print cluster summary Co-authored-by: ck-intel <chandrakant.khandelwal@intel.com> * Fetch and print OpenVINO version that OVTF was built against (#142) * Fetch and print OpenVINO version that OVTF was built against * Add import openvino_tensorflow to TF tests using find and replace instead of using patch (#141) * Enable cluster fallback through static initializers; Change default summary log info (#146) * Enable cluster fallback through static initializers; Change default summary log info * OpenVINO 2021 4 Upgrade Docs, CI, and Tests (#145) * Update OV version in docs, pipelines, and code * Update UT lists; fix OpenVINO 2021.4 download links in Dockerfiles * Fix ci random failures (#148) * Documentation review changes * Remove Code format check from Ubuntu-20 as clang-format-3.9 is not available in it * Installation table updated with MAC OS entries (#153) * Enable macOS Support (#133) * Initial commit for macOS * Merge commit '03379a2b92e0e11e94c6e6f709e3ee7770723f1a' * add TF 2 dylib var * fix virtualenv creation in py3 * Changing default OpenVINO version 2021.3 due to SetBlob issue * identify python minor version for load_venv * Adding the mac azure pipelines for ovtf project * Updating mac yml files * Fix CPU plugin issue, Fix unit test spawn issue for macOS * add mkldnn lib for release and debug build types * Update mac azure pipeline * Update OV version in docs, pipelines, and code * Update darwin tests list * Rebased on pre_gold_dev; Remove build errors; Update tests list * Enable VPU in OpenVINO, make on all cores * Update ABI0 based build pipeline on macOS * Add rpath to myriadPlugin and include myriad unit tests * Add myriad UT list files * Run myriad TF UT directly through tf_unittest_runner due to mutex error * Use -march=native only for Linux Co-authored-by: suryasidd <surya.siddharth.pemmaraju@intel.com> Co-authored-by: bhadur <bhadur.a.sm@intel.com> * Supporting backends (#152) * GPU_FP16 changes (#149) * Device name check for GPU FP16 Co-authored-by: suryasidd <surya.siddharth.pemmaraju@intel.com> * Export IR API is and related functionality is added (#150) * Document text correction for export ir * Fix TF1.x build failures that were due to Timestamp logic and dtype errors (#155) * Conditional compilation for timestamp logic; Fix DType error in TF 1.x in ovtf_builder * Models list updated (#151) * Corrected link for NCF-1B model * Devices name alignment in table Co-authored-by: Chandrakant Khandelwal <chandrakant.khandelwal@intel.com> * Fixed cmake flags issue and updated OV branch tags for previous released versions (#156) * Upgrade OVTF Python Samples to TF 2.x (#135) * added multi input support (image, video, camera) * added python examples to cpu ci pipline * removed tf 1.15.2 from req * api changes to tf2.5 * Python Inference Example Detection stage update * removed python samples in ci stages * tested camera input * Adding TF1x samples folder * TF 1.x yolo v3 model implementation to TF 2.x * added the pip packges to requirements * added assertion for input_file * added python sample to ci pipeline * added matplotlib as requirement * added TF1 python sample to ci pipeline * added requirements.txt * added --input arg help message * removed matplotlib installation * added requirements.txt to python TF2 stage * removed pip packages for ovtf samples * added python3.6 for TF1.x yolov3 conversion * bug fix for TF1.x python sample on MYRIAD * explicitly added the backends * python version check * installing python3.6 * saving the .h5 keras format for yolov3 darknet * added TF2.x API inference on .h5 kears format * added TF2.X APIs for inference with .h5 model * rollback to tf.compat.v1 for loading the graph * Renamed to TF_1_x and removed the unused files. * updated the TF1.x folder name to TF_1_x * removed filtering for listing ovtf backends * removing unused files * renaming output model * removed data folder in TF_1_x , using common utils * default model name change * Updated help message * removed python TF1 OD samples * assigned 0 for camera input * updated CI pipelines for python samples * assigned 0 for camera input * Unit tests update (#116) * TF unit test updates for all the devices * Add Explicit Padding to Conv2D. * Update TF unit test cases for MAC OS Co-authored-by: Peetha Veeramalai <preetha.veeramalai@intel.com> * OVTF Mandarin documentation (#138) * Updated OVTF Mandarin documentation Co-authored-by: Chandrakant Khandelwal <chandrakant.khandelwal@intel.com> * GPU core dump issue fixed (#159) Co-authored-by: Mustafa Cavus <cavusmustafa@intel.com> * Updated Mandarin readme documentation (#162) * Samples README update and yolov3 anchors file addition (#157) * added yolov3 anchors file * added license and writing output detections * changeed python default version to python3 * releasing memory after backends_len call * check backend updated and removed MIT license Co-authored-by: pratiksha123507 <pratikshax.bapusaheb.vanse@intel.com> * Update documentation for BUILD.md, INSTALL.md, and add requirements.txt (#158) * Add dependency install steps for Ubuntu and macOS * Add Bazelisk optional step * Upgrade OVTF version across files * Add separate instruction for PIP installations and update docs * Fix networkx version as 2.5.1 is common supported version for both Ubuntu 18.04 and 20.04 * Updated the installation commands in Interactive Table * Change PIL minimum required version * Added INSTALL.md link * Updated the build dependencies steps Co-authored-by: chandrakant khandelwal <chandrakant.khandelwal@intel.com> Co-authored-by: Ritesh Rajore <ritesh.kumar.rajore@intel.com> * Updated readme and license; Update interactive installation table (#163) Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * ManyLinux2014 Wheel Generation (#165) * Add patch file for manylinux * Update manylinux2014 Dockerfile * Add python 3.9 * Update versions in main.js * Dynamic fallback output fix (#166) * Dynamic fallback output issue fixed * Added INT8 quantization support (#144) * Added Translations and Transposes needed for Int8 Quantization * Added documentation for INT8 quantization support * Added CPU device condition for Relu6 and Constant Folding pass * Updated the CPU condition * Removed default enabling of constant folding on CPU Co-authored-by: chandrakant khandelwal <chandrakant.khandelwal@intel.com> * Version number changes in docs, code, and pipelines (#169) * updated MODELS.md (#167) * Upgrade to OV 2021.4.1 version (#170) * Code changes for OV 2021.4.1 upgrade * Updated docs for 2021.4.1 * Add changes to ABI1 whl generation Dockerfile Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Updated docs based on review comments (#171) * Document updates for GPU_FP16 * Update TROUBLESHOOTING.md * Additional document updates on GPU_FP16 * Update AWS_instructions.md * Update Azure_instructions.md * Installation table intro modified Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com> Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> Co-authored-by: Sai Jayanthi <sai.jayanthi@intel.com> * TF version updated in colab notebooks & Update examples documentation (#168) Co-authored-by: adisan3 <adityax.sanivarapu@intel.com> * Update OCM submodule(#172) * Export IR race condition issues are removed (#173) * Pillow version update (#175) * Updating readme for users using OVTF with CUDA capable GPUs (#177) * Model doc files updated with new models (#178) * Model doc files updated with 2 new models * Added complete model support for Facenet and lm_1b * Added warm-up inference run for samples (#179) * Added warm-up for first inference * Add Python3.9 Support for ABI1 Wheels (#176) * Add support for Python 3.9 wheels in ABI1 builds * Add python3.9-dev apt package * Updated INT8 quantization documentation in MODELS.md (#181) * Desciprtion for enabling oneDNN added to the README.md (#182) * Updated MODELS_cn doc (#180) * Update PyPi Documentation (#183) * Updated models file (#186) Co-authored-by: Chandrakant Khandelwal <chandrakant.khandelwal@intel.com> * Update Mandarin Documentation and DevCloud Links(#185) * Add updated Mandarin documentation Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> * Update mandarin readme and models file (#187) * Print TF version instead of git version (#188) Co-authored-by: Yamini Nimmagadda <yamini.nimmagadda@intel.com> Co-authored-by: Cavus Mustafa <mustafa.cavus@intel.com> Co-authored-by: mohdansx <mohdx.ansari@intel.com> Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com> Co-authored-by: Sai Jayanthi <sai.jayanthi@intel.com> Co-authored-by: adisan3 <adityax.sanivarapu@intel.com> Co-authored-by: bhadur <bhadur.a.sm@intel.com> Co-authored-by: Ambarish Das <ambarish.das@intel.com> Co-authored-by: Surya Siddharth Pemmaraju <surya.siddharth.pemmaraju@intel.com> Co-authored-by: pratiksha123507 <pratikshax.bapusaheb.vanse@intel.com> Co-authored-by: Peetha Veeramalai <preetha.veeramalai@intel.com> Co-authored-by: Mustafa Cavus <cavusmustafa@intel.com> Co-authored-by: Ritesh Rajore <ritesh.kumar.rajore@intel.com>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.