saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-22 22:01:08 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
2182 commits 2856 branches 57 tags 1.3 GiB
Commit graph

2182 commits

This branch
This branch
All branches
Author SHA1 Message Date
Jeff Bloomfield
8cc161aec6 Remove problematic change for dxcore.lib 2020-04-25 18:48:07 -07:00
Jeff Bloomfield
c49cc0c937 Increase DML nuget version to 0.0.2 2020-04-25 16:28:19 -07:00
Dwayne Robinson
dd86e3be10 Merged PR 4596882: Fix assert in ReadbackFromGpu 
`ExecutionProviderImpl::CopyTensors` calls `ReadbackFromGpu`, and depending on the mix of source/destination tensors (CPU/CPU, CPU/GPU, GPU/CPU, GPU/GPU), there can be anywhere from 0 to multiple tensors to copy. Copying 0 tensors is not an assertable failure and should just be a nop (tests work fine in release build). This assert reproes locally when running Pad tests (and some other operator tests too).
 2020-04-24 01:12:50 +00:00
Jeff Bloomfield
7c0b05eca0 Merged PR 4593169: Handle empty tensors in DML EP batched tensor copy 
This fixes the test_slice_start_out_of_bounds ONNX test

Related work items: #24837735
 2020-04-22 23:59:09 +00:00
Jeff Bloomfield
ab2b85649e Merged PR 4587771: Merge Github master to DmlDev 
This merges ORT Github master from to ADO DmlDev, in preparation with merging back to Github for ORT 1.3.

This branch includes changes from the following PRs, which will now be closed:
https://github.com/microsoft/onnxruntime/pull/3533
https://github.com/microsoft/onnxruntime/pull/3529
https://microsoft.visualstudio.com/WindowsAI/_git/onnxruntime/pullrequest/4569080

Related work items: #25935271
 2020-04-22 21:58:01 +00:00
Dwayne Robinson
26282359bf Merged PR 4591959: Fix ORT DML EP's Slice shape operator helper 
Related work items: #24672220
 2020-04-22 21:55:46 +00:00
Jeff Bloomfield
c2a01b9431 Disable erroneous compiler warning in space_depth_ops.cc 2020-04-21 01:40:12 -07:00
Jeff Bloomfield
971b98f9a5 Fix ARM build error 2020-04-20 17:15:55 -07:00
Jeff Bloomfield
850ab19e62 Fix Winml test build error 2020-04-20 15:31:16 -07:00
Jeff Bloomfield
19cdd6f1e1 Fix chk build error 2020-04-20 11:34:07 -07:00
Jeff Bloomfield
5d2874298e Merge remote-tracking branch 'upstream/user/jeffbloo/FreeDimOverrideByName' into user/jeffbloo/MergeGithubMasterToDmlDevPlusPending 2020-04-19 13:50:21 -07:00
Jeff Bloomfield
88732cd092 upstream/jeffbloo/TrimOnSessionInitializationEnd 2020-04-19 13:49:23 -07:00
Jeff Bloomfield
eceb18869a Merge remote-tracking branch 'origin/user/jeffbloo/BatchTensorCopy' into user/jeffbloo/MergeGithubMasterToDmlDevPlusPending 2020-04-19 13:45:31 -07:00
Jeff Bloomfield
acbfa42647 Merge remote-tracking branch 'origin/DmlDev' into user/jeffbloo/MergeGithubMasterToDmlDevPlusPending 2020-04-19 13:44:25 -07:00
Jeff
7d523d2580 Merge remote-tracking branch 'upstream/master' into jeffbloo/TrimOnSessionInitializationEnd 2020-04-19 11:58:44 -07:00
Jeff
414c4174a4 Merge remote-tracking branch 'upstream/master' into user/jeffbloo/FreeDimOverrideByName 2020-04-19 11:57:42 -07:00
Jeff Bloomfield
8ee5953153 Merge remote-tracking branch 'upstream/master' into user/jeffbloo/MergeGithubMasterToDmlDev1 2020-04-19 11:52:44 -07:00
Jeff Bloomfield
a4e312da43 Fix build error in D3DDeviceCache.cpp 2020-04-19 11:52:41 -07:00
Zhang Lei
422266c445
Support conv transpos 1D in cuda provider. (#3300) 
* Support conv transpos 1D in cuda provider.

* Clear some old comment. Enable conv_transpose_1d onnx test for cuda.
 2020-04-19 22:07:34 +08:00
Scott McKay
7d5348f87e
Add ability to batch device copy for graph inputs and outputs. (#3580) 
* Add ability to batch device copy for graph inputs and outputs.
 2020-04-19 17:51:07 +10:00
Prabhat
ea62b3435a
Clean up build.py code (#3466) 2020-04-18 20:48:30 -07:00
Maxim Kalinin
fcf0f6ee9f
Generalize reshape fusion (#3554) 
* Generalize reshape fusion

* Allow arbitrary number of Concat arguments
* Apply fusion even when an output of an internal node is used elsewhere
* Fix a bug when an internal node's output is the subgraph output
* Simplify code
 2020-04-18 20:47:23 -07:00
Tiago Koji Castro Shibata
14e387aa1a
Fix WinML namespace build break (#3583) 
* Add missing winrt namespace

* Conditional compilation of dxcore code

* Fix TAEF macros
 2020-04-18 20:46:01 -07:00
Sherlock
56b223bc60
Implement OneHot CUDA Kernels (#3390) 
* Implement OneHot CUDA Kernels

* Support fp16

* Use HandleNegativeAxis

* Make MLFloat16 test GPU only
 2020-04-18 17:41:39 -07:00
Hariharan Seshadri
1599562016 Fix BatchNorm CUDA kernel definition 2020-04-18 17:21:29 -07:00
Zhang Lei
c365822808
Refactor some for the calibate.py. Add QLinearAdd and QLinearMul support. Fix bugs loading jpgs not strict RGB, and typoes in load_batch call. (#3542) 2020-04-18 17:10:55 -07:00
Dmitri Smirnov
db9566f70d
Implement Inverse(12) for CPU and CUDA (#3485) 2020-04-18 17:10:21 -07:00
Dmitri Smirnov
38a18023c7
Fix some too popular warnings. (#3578) 
Some pointless and noisy warnings either fixed or disabled.
 2020-04-18 17:05:05 -07:00
Changming Sun
d68245853e
Disable downloading test data on Linux (#3581) 2020-04-18 15:54:58 -07:00
Sergii Dymchenko
3e884b4b6b
Fix some typos. (#3582) 
* Fix some typos.

* Fix a typo.
 2020-04-18 14:18:05 -07:00
suryasidd
6fe688c732
Disabled failed maxpool test on GPU (#3549) 2020-04-18 13:49:42 -07:00
Nick Feeney
e2288ff2b4 Merged PR 4574388: Implement Eyelike int ops 
Implement Eyelike int ops

Related work items: #24673994
 2020-04-18 01:11:31 +00:00
Dwayne Robinson
dc576a8de8 Merged PR 4574316: Pad, OneHot, DepthToSpace, SpaceToDepth, TopK, Where int registrations 
Related work items: #24673980, #24674011, #24674018, #24674032, #24674039
 2020-04-18 01:05:33 +00:00
Jeff
c47490ab31 Bug fix 2020-04-17 17:16:14 -07:00
Nick Feeney
c170d087a1 Merged PR 4573751: Add int32 and uint32 support to onnx Elementwise math 
Add int32 and uint32 support to onnx Elementwise math to make sure we don't diverge from onnx for new TF op.

Related work items: #26023998
 2020-04-17 23:33:50 +00:00
Tianlei Wu
7f46f347db
Add GPT2 Attention Fusion in optimization script (#3488) 
* Add Attention fusion for GPT2
* Support distilgpt2 in benchmark_gpt2.py
* Add options to disable Attention/SkipLayerNormalization/EmbedLayerNormalization/BiasGelu fusions
* Add logging at the begining of each fusion
* Update notebooks: Add Gpt2OnnxModel.py to list of script files.
* Add test for gpt2 model optimization
* Add optional parameters (--input_ids --segment_ids --input_mask) for graph inputs
* Fuse BiasGelu
* Handle model that does not have segment_ids input.
* Allow fuse embed layer without mask
 2020-04-17 16:23:53 -07:00
Tianlei Wu
5d3b217039
Update Attention operator for GPT2 (#3474) 
Add unidrectional mask for Attention operator.
Update mask_index to mask broadcast from B->BxS->BxNxSxS to B->BxSxS->BxNxSxS.
 2020-04-17 16:20:40 -07:00
Jeff
027b0cb3f3 Update to match ORT signature 2020-04-17 15:14:01 -07:00
Hariharan Seshadri
b4457ecb7a
Fix gen_doc build option and refresh documentation (#3545) 
* Support listing keys in custom metadata map via C/C++ API

* nit

* PR feedback

* Nit

* Initial commit

* More changes

* Support listing keys in custom metadata map via C/C++ API

* nit

* PR feedback

* Nit

* Initial commit

* More changes

* Add md files

* Doc changes

* Update

* revert cmake changes

* Update

* Doc change

* Update

* Update
 2020-04-17 14:41:04 -07:00
Hector Li
5acd8dbe7d
remove option --enable_lto (#3515) 2020-04-17 14:18:56 -07:00
Jeff Bloomfield
11baa4b56d Merge user/ticastro/fix-wai-build 2020-04-17 14:10:50 -07:00
Jeff Bloomfield
822883314e merge upstream/master 2020-04-17 13:47:07 -07:00
Yufeng Li
f822a54860
Make De/QuantizeLinear support half (#3531) 
* Make QuantizeLinear support half

* remove unnessary type constraint

* refine kernel definition

* add fp16 support for dequantizelinear

* diable QuantizeLinear_per_tensor_half_int8 for tensorrt

* refine unit test and fix saturate issue for MSDomain QuantizeLinear

* fix build break

* include tensorrt for half_uint8 test
 2020-04-17 12:17:48 -07:00
Tracy Sharpe
c7b6fab29d
Fix build break in mlas\lib\quantize.cpp: missing nearbyintf (#3572) 2020-04-17 11:50:25 -07:00
Nick Feeney
eda4df62ab Merged PR 4569464: ArgMin ArgMax Integer implementation 
ArgMin ArgMax Integer implementation

Related work items: #24673965, #24673969
 2020-04-17 18:00:57 +00:00
Xiang Zhang
43c3a5edba
update onnxruntime version string for telemetry (#3526) 
* update onnxruntime version string for telemetry

* use ORT_VERSION

* deleted version.h
 2020-04-17 10:46:58 -07:00
Changming Sun
209b41a67d Update dependencies graph 2020-04-17 07:38:45 -07:00
Sheil Kumar
2717c178cc
Fork the WinML APIs into the Microsoft namespace (#3503) 
* Migrate winml to Microsoft Namespace (packaging changes are pending)

* add ns_prefix toggle

* fix packaging

* Users/sheilk/add missing raw header (#3484)

* add dualapipartition

* wrong variable for repo root

Co-authored-by: Sheil Kumar <sheilk@microsoft.com>

* remove existence check to force failures

* extra paren

* dualapipartition needs to be referenced from the source

* add microsoft.ai.machinelearning.dll to the output dir

* rename the idl file so that assembly info is correctly added into the winmd

* fix namespaces

* update namespaces

* default to microsoft, and add namespace override as build argument

* update cmakesetings.json as well

* remove from cmakelists.txt

Co-authored-by: Sheil Kumar <sheilk@microsoft.com>
Co-authored-by: Changming Sun <chasun@microsoft.com>
 2020-04-17 06:18:54 -07:00
ytaous
fcb27c4e8b
hotfix for skiplayernorm (#3543) 
Co-authored-by: Ethan Tao <ettao@microsoft.com>
Co-authored-by: Changming Sun <chasun@microsoft.com>
 2020-04-17 01:22:08 -07:00
liuziyue
92269ae409
perf tuning docs update (#3520) 2020-04-17 00:23:15 -07:00