saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-31 23:27:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
6493 commits 2856 branches 57 tags 1.3 GiB
Commit graph

6493 commits

This branch
This branch
All branches
Author SHA1 Message Date
Changming Sun
2d2eebb844 Correct a comment 
"WINVER=0x0602" means  Windows 8.  source: https://docs.microsoft.com/en-us/cpp/porting/modifying-winver-and-win32-winnt?view=msvc-170
 2022-03-11 11:42:41 -08:00
Ryan Lai
2e7592ddf8
avoid using LocalFree on FormatMessageW buffer (#10796) 
* remove local free

* Remove local free from onnxruntime

* don't allocate

* Change to use constexpr to satisfy  CPU build warning
 2022-03-11 11:11:40 -08:00
Kotaro Yamamoto
64556888a1
add python binding for RunOptions config entry (#10694) 2022-03-11 08:49:22 -08:00
pengwa
d478a53d43
don't clear grad_fns & add test (#10671) 2022-03-11 14:31:54 +08:00
Edward Chen
1a62306db7
Use separate build directories for full and mobile iOS packages. (#10835) 2022-03-10 19:33:06 -08:00
Chun-Wei Chen
5202efd11e
remove unused six in code and CIs (#10832) 2022-03-10 15:38:44 -08:00
Changming Sun
f87a06cd96
Patch absl so that it doesn't disable important VC++ warnings (#10836) 
This PR is just for making onnxruntime passing Binskim rules.

Below is how I made it:

git clone absl repo, checkout the version we are using
Then apply our patch file
Make modifications
Regenerate the patch file by "git diff > C:\src\onnxruntime\cmake\patch\xxx.patch"
Then submit the change to our repo
You will need to repeat the steps when you need to advance the absl commit or add more changes to it.
 2022-03-10 15:35:39 -08:00
Pranav Sharma
97ae44d060
Mark end of version 11 C API. (#10803) 
* Mark end of version 11 C API

* Add static_assert
 2022-03-10 15:11:02 -08:00
Abhishek Jindal
3ae2bfaefe
Abjindal/torch api change gelu (#10833) 
* changing gelu backward op and adding required files

* cleaning up file and adding comments

* version comparison issue
 2022-03-10 11:56:30 -08:00
Dmitri Smirnov
1d545dfe87
Address performance issue with abseil flat_hash_table. (#10819) 
When returning by value in a cross DLL call, the hash table
even though containing all the entries that are originally there
can not find at least some of them. Reverting to std::unordered_set
pending further investigation.
 2022-03-10 09:49:55 -08:00
Hariharan Seshadri
e80ff63274
Fix bug in MemcpyToHost (#10816) 2022-03-10 07:02:27 -08:00
Ryan Hill
9853eaa14f
Detect runtime CUDA JIT and warn the user (#10781) 
* Use cudaMalloc vs cudaDeviceSynchronize and show the total time
 2022-03-09 19:15:16 -08:00
Changming Sun
cc3a3476ed
Uninstall onnxruntime-training before running local tests (#10827) 
* Uninstall onnxruntime-training before running local tests
 2022-03-09 18:45:04 -08:00
zhangyaobit
9cbcc93e03
Add micro-benchmarks for Attention and SkipLayerNormalization ops. (#10798) 
* Add micro-benchmarks for Attention and SkipLayerNormalization ops.

* Add choices for argument provider and precision.

* Automatically select CUDA or ROCM execution provider.
 2022-03-09 18:18:51 -08:00
Abhishek Jindal
1c313f4476
changing gelu backward op and adding required files (#10813) 
* changing gelu backward op and adding required files

* cleaning up file and adding comments
 2022-03-09 16:54:51 -08:00
Edward Chen
0293e525ea
Make QDQSelectorActionTransformer() is_int8_allowed parameter required. (#10820) 
Make QDQSelectorActionTransformer() is_int8_allowed parameter required.
Set it to QDQIsInt8Allowed() in places it was previously set to false.
 2022-03-09 16:19:43 -08:00
Changming Sun
cc6bc34c8c
Update protobuf submodule (#10801) 2022-03-09 09:37:58 -08:00
Dmitri Smirnov
58521fb822
Make training CUDA kernels to adhere established code structure patterns (#10735) 
Current training optimizer kernels include CPU headers
  that affects changes that we can make in the CPU code with C++14 compiler and
  other refactoring efforts. Rearrange the kernel according to the established patterns
  and do not include headers that are not needed.
 2022-03-09 09:06:45 -08:00
Adam Pocock
4ef81b142d
Making the Java tests faster by optionally disabling ones which require running multiple JVMs. (#10811) 2022-03-08 22:19:37 -08:00
Hariharan Seshadri
ae97ecf05b
Fix CPU, CUDA Selu activation logic (#10771) 2022-03-08 19:53:27 -08:00
Edward Chen
c147c9dda6
Remove ORT_ENABLE_RUNTIME_OPTIMIZATION_IN_MINIMAL_BUILD. (#10778) 
Remove ORT_ENABLE_RUNTIME_OPTIMIZATION_IN_MINIMAL_BUILD as it is now implied by ORT_EXTENDED_MINIMAL_BUILD.
Remove related CMake option.
 2022-03-08 16:18:49 -08:00
George Wu
769aa8363d
update onnx-tensorrt to bring in https://github.com/onnx/onnx-tensorrt/pull/812 (#10810) 2022-03-08 14:51:07 -08:00
Jingqiao Fu
f4fd67cc2c
Revert "add load from buffer (#10162)" (#10590) 
This reverts commit 5cd57bb726.
 2022-03-08 13:35:23 -08:00
dependabot[bot]
7e04dccca7
Bump numpy in /tools/ci_build/github/linux/docker/scripts (#10385) 
Bumps [numpy](https://github.com/numpy/numpy) from 1.16.6 to 1.21.0.
- [Release notes](https://github.com/numpy/numpy/releases)
- [Changelog](https://github.com/numpy/numpy/blob/main/doc/HOWTO_RELEASE.rst.txt)
- [Commits](https://github.com/numpy/numpy/compare/v1.16.6...v1.21.0)

---
updated-dependencies:
- dependency-name: numpy
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2022-03-08 11:02:36 -08:00
Sunghoon
68c8f5a1ef
Change a pipeline vmImage from windows-latest to windows-2019 (#10804) 2022-03-08 10:49:59 -08:00
Yufeng Li
33c6819196
add qdq support of Sigmoid (#10800) 2022-03-08 10:29:15 -08:00
Changming Sun
6260733533
Fix eager mode pipeline (#10802) 
It was still using python 3.6
 2022-03-08 09:26:20 -08:00
Hariharan Seshadri
a9d9c6b486
Register CPU, CUDA and ROCM opset-16 kernels for some operators (#10643) 2022-03-08 09:18:39 -08:00
Changming Sun
ce07dc30fd
Change how we apply patches to absl (#10799) 2022-03-08 02:03:06 -08:00
George Wu
1e4a4bfe58
update onnx-tensorrt reference. (#10795) 2022-03-07 21:45:46 -08:00
liqun Fu
da885a72e8
update with onnx 1.11 release (#10441) 2022-03-07 21:10:55 -08:00
Yulong Wang
80917342b7
[js] upgrade mocha@8.2.1 to 9.2.1 (#10793) 2022-03-07 20:40:24 -08:00
dependabot[bot]
4d943c9bd3
Bump numpy from 1.16.6 to 1.21.0 in /tools/ci_build/github/linux/docker/scripts/manylinux (#10387) 
* Bump numpy in /tools/ci_build/github/linux/docker/scripts/manylinux
 2022-03-07 20:39:49 -08:00
PeixuanZuo
c07a27a008
[FIX] delete python3.6 from AMD python package docker image builder (#10790) 
* [UPDATE] delete python3.6 to cooperate numpy==1.21.0

* [UPDATE] delete python3.6 to cooperate numpy==1.21.0
 2022-03-07 18:21:43 -08:00
Vincent Wang
4a38f9e31d
enable strided tensor for training only (#10748) 2022-03-08 08:31:28 +08:00
zhangyaobit
b7f00b9682
Refactor the common code per operator into an abstract base class. (#10785) 2022-03-07 13:15:49 -08:00
Daigo HIROOKA
a08036da09
correct symbolic name of GridSample operation (#10782) 
Function name needs to match PyTorch ATen op name, which is `aten::grid_sampler`.
 2022-03-07 12:49:12 -08:00
dependabot[bot]
3e54f94bb0 Bump karma from 6.3.14 to 6.3.16 in /js/web 
Bumps [karma](https://github.com/karma-runner/karma) from 6.3.14 to 6.3.16.
- [Release notes](https://github.com/karma-runner/karma/releases)
- [Changelog](https://github.com/karma-runner/karma/blob/master/CHANGELOG.md)
- [Commits](https://github.com/karma-runner/karma/compare/v6.3.14...v6.3.16)

---
updated-dependencies:
- dependency-name: karma
  dependency-type: direct:development
...

Signed-off-by: dependabot[bot] <support@github.com>
 2022-03-07 11:47:23 -08:00
Yulong Wang
25fdcfbd14
[js/web] allow multiple inference session creating concurrently (#10784) 
* test case

* bugfix

* fix

* support multi session init
 2022-03-07 11:35:06 -08:00
RandySheriffH
a4b5fa334a
Add type and shape information to profiled numbers (#10773) 
* add func to collect type shape

* reformat

* refactor perf view

* remove obsolete
 2022-03-07 10:17:58 -08:00
Changming Sun
d8bf9a479b
Remove python 3.6 from training pipelines (#10780) 
Because the numpy we use doesn't support python 3.6. And inference pipelines already removed python 3.6.
 2022-03-07 09:57:24 -08:00
Hariharan Seshadri
9d30262422
Fix AMD training pipeline (#10788) 2022-03-07 08:53:08 -08:00
Chen Fu
50a6f095cd
Symmetric QGEMM kernel for ARMv8 A55 chip (#10754) 
ARM a55 micro-architecture (with dot product instructions), similar to a53, is widely used as little cores in big.Little configurations. A55 has a narrower memory load/store hardware, where a 128b load instruction would block the pipeline for 2 whole cycles, during which no other instructions can be executed. On the other hand, a 64b load instruction can be duo issued with many other instructions.

This change adds a Symmetric QGEMM kernel for a55 micro-architecture, where we replace

ldr q4,[x1],#16

with

ldr d4,[x1],#8
ldr x11,[x1],#8
ins v4.d[1],x11

so that we can try to hide the memory load cycles behind computing cycles in the kernel.

Co-authored-by: Chen Fu <fuchen@microsoft.com>
 2022-03-07 08:41:13 -08:00
PeixuanZuo
55af7a96a7
update the amd ci pipeline (#10723) 
* [TEST] test to get amd pipeline information

* [FIX] lower the threshold

* [UPDATE] add retry task

* [UPDATE] add retry task

* [ERROR] error to occur retry

* [FIX] error

* [UPDATE] update retryCountOnTaskFailure to 1 time

* [UPDATE] add showmeminfo
 2022-03-07 18:39:42 +08:00
Fei Hu
60acfd3dd8
Support CUDA Graph in the CUDA EP (#9978) 2022-03-06 20:47:31 -08:00
Tianlei Wu
0e335aba37
Update BeamSearch operator spec to support t5 (#10777) 
* change BeamSearch op to support encoder decoder model

* check model_type and decoder attribute

* fix

* update comments

* warn shape inference issue with onnx v1.11 or T5

* skip parity test when tempature != 1.0

* fix build
 2022-03-04 21:52:45 -08:00
George Nash
6be5185088
Update dnnl Add, Mul, Sub, Div ops to handle scalar values (#10756) 
* Update dnnl Add, Mul, Sub, Div ops to handle scalar values

Signed-off-by: George Nash <george.nash@intel.com>

* Add additional scalar support for dnnl execution provider

This will add scalar support for:
Eltwise operators: Abs, Elu, Exp, LeakyRelu, Log, Relu, Round,
                   Sigmoid, Softplus, Sqrt, and Tanh
Gelu operators: BiasGelu, FastGelu, and Gelu
Softmax operator

Signed-off-by: George Nash <george.nash@intel.com>
 2022-03-04 19:28:25 -08:00
Ye Wang
259ade2557
Add ability to modify num_hidden_layers from benchmark script (#10760) 
* add ability to modify num_hidden_layers from benchmark script

* comment

* Revert "comment"

This reverts commit 28794b0e4f86506dcc937738894fcef97fc84e48.

* Revert "add ability to modify num_hidden_layers from benchmark script"

This reverts commit 96f36ed7f751721bcf4e3ab8748a715f19a4e044.

* review coments

Co-authored-by: Ubuntu <wy@linux-v100.aidmrjtolptuzevavgwhrapqcd.jx.internal.cloudapp.net>
 2022-03-04 18:28:51 -08:00
Ella Charlaix
fde847473b
Add min max moving average calibration method (#10753) 
* Add min max moving average calibration method

* Modify the calibration extra options dictionnary creation
 2022-03-04 14:55:31 -08:00
Maxiwell
43ff27c7c8
ppc64le: optimizing the MlasQuantizeLinear() with VSX (#10644) 
This code is valid only when -mcpu is set to utilize POWER9 technology
or above. A compatible code for POWER8 was created as well, but it
was not tuned for performance.
 2022-03-04 14:54:56 -08:00