saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-23 22:13:38 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
6051 commits 2856 branches 57 tags 1.3 GiB
Commit graph

6051 commits

This branch
This branch
All branches
Author SHA1 Message Date
Sherlock
175acf08f4
ScatterND supports negative indices (#9739) 
* ScatterND supports negative indices
 2021-11-30 21:17:32 -08:00
Hariharan Seshadri
c161813217
Misc InstanceNorm CUDA kernel changes (#9879) 2021-11-30 19:15:45 -08:00
Sherlock
6de79d82c8
Fix Training Packaging pipeline (#9885) 
* Fix Training Packaging pipeline
 2021-11-30 15:26:10 -08:00
Abhishek Jindal
740679d329
Abjindal/fix windows ci pipeline (#9883) 
* switching to /wd4800 for eager mode

* fixing compile flags ignore warnings, previously it was only using the last one
 2021-11-30 10:33:13 -08:00
RandySheriffH
9345894c82
Add build option to enable cuda profiling (#9875) 2021-11-29 22:44:50 -08:00
RandySheriffH
16bfd3c771
Cancel transpose optimizer for resize (#9870) 
* cancel transpose optimizer for resize

* add UT

* addressing comments

* fix build err
 2021-11-29 22:30:16 -08:00
Chi Lo
0baf687f2d
QDQ tool modification part2 (#9720) 
* Add finetuned qdq options

* Add description

* Add unit tests

* Modify for channel axis

* Remove too specific feature. Move this implementation to e2e example

* Add OpTypesSupportPerChannelQuantization

* fix bug for unit test

* Keep flags OpTypesSupportPerChannelQuantization and QDQChannelAxis for internal use 

Will have a follow-up PR to fine tune the code

* remove unnecessary warning

Co-authored-by: stevenlix <38092805+stevenlix@users.noreply.github.com>
Co-authored-by: Yufeng Li <liyufeng1987@gmail.com>
 2021-11-29 21:45:41 -08:00
Ryan Lai
57a6f7c205
Various fixes to fix WindowsAI RI build. (#9877) 
* WAI RI fixes

* span changes

* Spaces

* Additional warnings to fix

* Fix redundant commment
 2021-11-29 21:33:15 -08:00
Yufeng Li
a0afd7303d
add int8_t support for pool operators (#9852) 
* add int8_t support for pool operators
 2021-11-29 18:43:43 -08:00
George Wu
ec9b0ed800
[python manylinux package] emit warning if missing CUDA/TensorRT dependency causes ld_preload to fail and user tries to register either CUDA/TensorRT EP (#9872) 
* add warning if ld_preload fails for CUDA or TRT when trying to register either provider

* refactor

* change wording from register to create
 2021-11-29 16:04:23 -08:00
Yulong Wang
27e337ed7a
[js/node] npm audit fix (#9861) 2021-11-29 13:11:19 -08:00
Ye Wang
bf716e667c
layernorm throw error if input has no data (#9837) 2021-11-29 12:43:17 -08:00
cristei
9e75ebf0dc
Remove redundant inline specifiers, sync server IsLittleEndianOrder with runtime core (#9856) 
* Remove redundant inline

* Make server IsLittleEndianOrder represent runtime core endianness implementation, make the endianness check constexpr
 2021-11-29 08:32:16 -08:00
Tang, Cheng
37bf46eb19
support print in ort eager mode (#9825) 
* fix reshape implementation in eager mode

* test code

* update opgen script to support fallback to cpu

* enhance the eager backend to support torch cpu fallback

* add more testes

* disable the printensor test for now, as we need to erge a PR to pytorch first
 2021-11-29 08:03:57 -08:00
George Wu
1e9e57df3e
no fallback when enforcing explicit EP registration. (#9863) 
* no fallback when enforcing explicit EP registration.

* add explicit ep registrations for python.
 2021-11-25 07:26:51 -08:00
Yulong Wang
a3ebc5e082
[js/web] do not use nodejs type 'Buffer' in web (#9839) 
* [js/web] do not use nodejs type 'Buffer' in web

* resolve comments and validate tests

* remove 'Buffer' in test
 2021-11-24 14:14:42 -08:00
Yulong Wang
6eb0c8d420
fix build break in release pipeline for Node.js binding test (#9850) 2021-11-24 13:54:43 -08:00
Sheil Kumar
53c43e9949
WinML RT API: Add PixelRange Metadata to Bind() call PropertySet (#9827) 
* Enable Normalization Binding Metadata

* copy paste error

* Small fix.

Co-authored-by: Sheil Kumar <sheilk@microsoft.com>
 2021-11-24 13:44:25 -08:00
Hariharan Seshadri
18fd2cf457
Fix potential data race with OrtValue usage in Python (#9841) 2021-11-23 14:34:24 -08:00
Maajid khan
0ae0f29f14
[OpenVINO-EP] V3.4 Release with OpenVINO 2021.4.2 LTS Release (#9848) 
* Changes to ensure openvino build go through in Windows

* Modified Hetero plugin Logic

*Modified Hetero Feature logic. In Hetero,
if the operator to be marked true in getcapability(),
it should be supported by either of the devices
specified with HETERO in the device_type.

Signed-off-by: MaajidKhan <n.maajidkhan@gmail.com>

* OV updated to 2021.4.2 version

* OV updated to 2021.4.2 version

* Updated OV to 2021.4.2 version, mono download  link and dotnet version

* Copying Managed nugets in openvino c# docker file

*Copying Managed nuget to nugets artifacts
directory

Signed-off-by: MaajidKhan <n.maajidkhan@gmail.com>

Co-authored-by: saharfraza <sfatima.3001@gmail.com>
Co-authored-by: mayavijx <mayax.vijayan@intel.com>
Co-authored-by: Aravind Gunda <aravindx.gunda@intel.com>
 2021-11-23 13:12:08 -08:00
Ginés Hidalgo
926109bd63
Added algorithm for std::clamp (#9797) 2021-11-23 12:58:05 -08:00
Dmitri Smirnov
d012d9fb91
Fix memset size (#9840) 2021-11-23 09:19:55 -08:00
Ryan Hill
6749e9fd44
Cuda instance_norm fix (#9826) 
* Fix allocation size & initial values
 2021-11-22 22:59:20 -08:00
Vincent Wang
24f3d72b77
relax atol and rtol for einsum ut (#9842) 2021-11-23 14:39:14 +08:00
RajalakshmiSR
8564fc1933
POWER10: Add optimized dgemm kernel (#9652) 
* POWER10: Add optimized dgemm kernel

This patch makes use of POWER10 matrix multiply assist feature and
adds new DGEMM kernel.

* Indentation update

Co-authored-by: Rajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
 2021-11-22 20:28:21 -08:00
jingyanwangms
bf5e9a5044
bumping up ORT_API_VERSION to 10 (#9838) 
Co-authored-by: Jingyan Wang <jingywa@OrtTrainingDev3.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2021-11-22 20:27:45 -08:00
Scott McKay
fb4a8e12fc
Limit inclusion of Xamarin mobile target frameworks. (#9834) 
- Only set them as targets for the ORT nuget package
    - Use OrtPackageId as the condition for inclusion, if installed
      - need to do the nuget restore via msbuild so that this property is set correctly
  - Add desktop-only version of the C# sln as there is no way to exclude the mobile specific csproj's from an sln
    - use this when applicable if someone is running build.py with the `--build_nuget` flag

Other
  - remove attempt to include symbols in the nuget package as nuget doesn't support symbols in native packages
  - update build.py to use `nuget` and not a windows specific path and filename for a linux build with `--build_nuget`
 2021-11-23 11:29:53 +10:00
Yulong Wang
74ca417c0e
[js/web] optimize bundle file size (#9817) 
* es2017 by default for ort-common

* add visualizer and define plugin

* es2017 for ort-web. also add build target for es5

* add multiple reduced size build for ort-web

* resolve comments, add e2e tests and add docs
 2021-11-22 13:56:55 -08:00
Edward Chen
bcc6ab29f6
Trim DataTypeImpl binary size (#9813) 
* De-virtualize DataTypeImpl::AsXType() functions.
* Refactor helpers.
 2021-11-22 12:06:24 -08:00
Dmitri Smirnov
567749b2dc
Expose IOBinding SynchronizeInputs/Outputs via C/C++/C# And Python APIs (#9823) 
Add C/C++ APIs for SynchronizeBoundInputs/Outputs
 Add python bindings
 Expose SynchronizeBoundInputs/Outputs to C# API
 2021-11-22 09:45:31 -08:00
dependabot[bot]
8dcd388ec5 Bump ansi-regex from 5.0.0 to 5.0.1 in /js 
Bumps [ansi-regex](https://github.com/chalk/ansi-regex) from 5.0.0 to 5.0.1.
- [Release notes](https://github.com/chalk/ansi-regex/releases)
- [Commits](https://github.com/chalk/ansi-regex/compare/v5.0.0...v5.0.1)

---
updated-dependencies:
- dependency-name: ansi-regex
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
 2021-11-22 16:04:10 +00:00
dependabot[bot]
c779f2cad5 Bump path-parse from 1.0.6 to 1.0.7 in /js 
Bumps [path-parse](https://github.com/jbgutierrez/path-parse) from 1.0.6 to 1.0.7.
- [Release notes](https://github.com/jbgutierrez/path-parse/releases)
- [Commits](https://github.com/jbgutierrez/path-parse/commits/v1.0.7)

---
updated-dependencies:
- dependency-name: path-parse
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
 2021-11-22 16:03:45 +00:00
Changming Sun
88d8d2d02a
Remove Copy NuGet Package to Azure Blob Store build step (#8991) 2021-11-22 08:03:02 -08:00
Yi-Hong Lyu
efcdbac4e7
Fix ONNX Runtime Mobile link in js/web/README.md (#9828) 2021-11-21 17:58:03 -08:00
Chi Lo
3a85ade511
Avoid segmentation fault when creating session with TRT EP or OpenVINO EP using python (#9814) 
* fix segmentation fault

* fix typo

* fix bug

* make logic the same as CUDA ep

* Modify for OpenVINO

* Add env variable check for OpenVIO

* refine the code

* refine EP failed registration warning messages.

* update OpenVINO exception message.

Co-authored-by: George Wu <jywu@microsoft.com>
 2021-11-20 07:52:50 -08:00
Dwayne Robinson
7396689c2b
Merge pull request #9765 from microsoft/user/dwayner/DML1.8forORT1.10 
Update DirectML 1.5.1 to 1.8.0 for ORT1.10
 2021-11-19 22:39:07 -08:00
Ye Wang
6856619b18
Decoder Attention CUDA Op (#9792) 
* add kernel interface

* register kernel

* add self/cross qkv projection without cache

* add LaunchTransQkv2 for (S,B,X,N,H) -> (X,B,N,S,H)

* refactor ConcatPastToPresent

* DecoderQkvToContext interface

* q,k,v buffer and cache as output

* qk, pv and transctx

* fix compiler error on linux machine

* key_padding_mask

* add test_parity file. However not runnable

* add partial unittest

* made partial attributes to inputs

* --gen_doc

* change kernel interface, add more tests

* morre parity tests

* fix test

* fix typo

* transpose optimizer has bug. remove it temporarily

* add input shape checks

* add type/shape inference

* fix cache shape check

* fix rocm build failure

* fix rocm build error

* review comments

* review comments
 2021-11-19 19:25:36 -08:00
Zhang Lei
16ddaf564c
Fix uninitialized warning by some compiler. (#9822) 
Co-authored-by: Lei Zhang <phill.zhang@gmail.com>
 2021-11-19 17:47:14 -08:00
Dwayne Robinson
f28d7eca67 Disable DML dynamicquantizelinear in backend tests DML Bug 33073263 2021-11-19 16:29:34 -08:00
Ryan Hill
1b953c6423
Fix some code defects (#9810) 2021-11-19 15:48:15 -08:00
Dwayne Robinson
d1b772e09a Mark dynamicquantizelinear as broken in DML 2021-11-19 15:06:44 -08:00
Sergii Dymchenko
ba339e667b
Add training performance investigation script (ONNX graph analyzer) (#9791) 
* Add first version of performance investigation script.

* Simplify and update performance investigation script.
 2021-11-19 13:27:00 -08:00
Dwayne Robinson
f047be55d4 Merge branch 'master' into user/dwayner/DML1.8forORT1.10 2021-11-19 12:42:52 -08:00
Zhang Lei
7e026286ae
Safe remove quantized inititalier which may still be used (#9788) 
* safe clear unused initializers.

* Force quantize by default.

* add option for force quantize.

* Add transpose to quantize_dynamic.

* optimize remove initialiers, catch error on remove input entries as
original logic.

* fix change impact on th qdq quantizer.

* Fix a typo

* add logic handle empty input/output name.
 2021-11-19 11:11:30 -08:00
Wei-Sheng Chin
e520bb5145
Improve print functions for NodeArg, Node, and Graph (#9801) 2021-11-19 09:48:27 -08:00
Gary Miguel
9d3c63263b
symbolic_shape_infer: Improve error message on mismatched types (#9809) 
The previous assertion failure was basically impossible to debug.
 2021-11-19 09:39:26 -08:00
Dwayne Robinson
289b1bdc86 Merge remote-tracking branch 'ado_wai_ort/DmlDev' into user/dwayner/DML1.8forORT1.10 2021-11-19 05:35:00 -08:00
Dwayne Robinson
32419974ad Merge remote-tracking branch 'origin/master' into user/dwayner/DML1.8forORT1.10 2021-11-19 05:20:26 -08:00
Dwayne Robinson
e0ffc30a0b Update to 1.8.0 2021-11-19 04:44:32 -08:00
Scott McKay
afd60a274c
Fix some places where there were unused parameters when sparse tensors were disabled. Doesn't break in an android/ios build. Does in a windows build. (#9807) 2021-11-19 14:17:53 +10:00