saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-27 22:45:57 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
1529 commits 2856 branches 57 tags 1.3 GiB
Commit graph

1529 commits

This branch
This branch
All branches
Author SHA1 Message Date
Yulong Wang
bf7fa091cc
NonMaxSuppression cuda implementation (#2082) 2019-10-31 11:53:22 -07:00
Changming Sun
67755adfd8 Bug Fix: NodeArg class has a move constructor but doesn't have a move assignment operator 2019-10-31 10:29:54 -07:00
RandySheriffH
d6849bd26c
Rashuai/cuda top k (#1919) 
* implement cuda topk

* implement heap

* add type support

* refactor interface

* add support for sorting by index

* add test case

* use cub device radix sort

* register for opset 9 and 10

* add opset 9/10 delaration

* refactor code

* refactor code

* fix comment

* fix comment

* switch to scratched mem
 2019-10-31 10:26:00 -07:00
Hariharan Seshadri
4bcd8bfca1
Fix CUDA Reduce ops (#2268) 
* Add some tests for Reduction ops

* Exclude tensorrt for new tests

* Fix bug in CUDA Reduce ops

* Fix nit
 2019-10-31 10:11:59 -07:00
Changming Sun
a5da5ff6f4 Remove onnxruntime_USE_EIGEN_THREADPOOL cmake option 2019-10-30 21:51:54 -07:00
KeDengMS
ff64d1f55b
Relax check for optimized model saving (#2291) 
So user may save model with layout optimization.
 2019-10-30 21:48:49 -07:00
Maik Riechert
ecfbb1bb99 Add missing guards to profiling calls (#1374) 
* guard remaining profiler calls

* enforce proper usage of profile class
 2019-10-31 14:28:49 +10:00
George Wu
aa041026e3
update Dockerfile.openvino (#2286) 
* install miniconda before openvino installation. add networkx, defusedxml dependencies.

* apt-get update

* apt-get update

* merge Intel changes.
 2019-10-30 13:58:24 -07:00
Tomasz Dołbniak
427e627805 Support for the Expand op with constant shape inputs (#2278) 
* Disable the Expand op for non-const shape inputs

* Check if an input is constant with IsConstantInitializer
 2019-10-30 13:22:45 -07:00
KeDengMS
e18c9582a8
[NupharEP] performance improvements (#2283) 
* [Nuphar EP] performance improvements
1. Add new ops: Shape, Expand
2. Add support for steps in Slice
3. Simplify Gather
4. Always inline alias nodes
5. Transpose nodes with inner loop being symbolic falls back to CPU provider when vectorization is not possible
6. Add opt_inproj option to model_editor to extract MatMuls inside Scan for input projection to outside
 2019-10-30 10:15:04 -07:00
zhijxu
63e9961637 fix typo 2019-10-30 09:57:56 -07:00
zhijxu
8dabe0502b merge two RUN to avoid making docker image too larger 2019-10-30 09:57:56 -07:00
Changming Sun
7b11f05a97 Update version number 2019-10-30 08:13:09 -07:00
Changming Sun
2172a9e5ed Fix an issue in the nuget run tests scripts 2019-10-30 08:13:09 -07:00
Changming Sun
138a7f194e Add cleanup step 2019-10-30 08:13:09 -07:00
Scott McKay
002d9c335b Add node and op type info to error message if there's a type or shape inferencing exception thrown by the ONNX checker (#2277) 
* Add node and op type info to error message if there's a type or shape inferencing exception thrown by the ONNX checker.

* Fix line break from auto format

* Remove unused param from unit test code.
 2019-10-29 23:42:13 -07:00
Ashwini Khade
8d231a32f2 Remove the libc version check in C# code (#2282) 2019-10-29 21:31:38 -07:00
Scott McKay
983a616bda
Revert to using opset 7 as the default for OpTester. Add explanation as to why that is: (#2256) 2019-10-30 09:42:21 +10:00
Scott McKay
47f40ca204
Cleanup add/remove of initializer (#2274) 
* Cleanup add/remove of initializer

* Address PR comments
 2019-10-30 06:50:41 +10:00
Dmitri Smirnov
09eb8ff8b8
make sure samples run (#2276) 
Adjust instructions for imagenet. Fix compile warning.
 Fix input/output names. Update build/run instructions.
 2019-10-29 10:07:35 -07:00
Changming Sun
d1096b524f Fix fns candy build error on Linux 2019-10-29 02:54:25 -07:00
Nathan
b0359b5622
register where kernel (#2245) 2019-10-28 15:59:29 -07:00
KeDengMS
5611a528f6
onnxruntime_perf_test: Add -u option to save optimized model (#2227) 2019-10-28 12:36:31 -07:00
Ryan Hill
3ecdd985cb
MNIST Sample Fix (#2259) 
* Fix Global variable initialization order

* Remove static initialization, and add error messages
 2019-10-28 11:22:45 -07:00
zhijxu
0b88eff43a add docker file to build onnxruntime with different execution providers 2019-10-28 10:03:18 -07:00
zhijxu
3926ce33f4 add --parallel to speed up compiling source code 2019-10-28 10:03:18 -07:00
zhijxu
4eda08534f make smaller step in Dockerfile 
so image layers can be reused as much as possible
 2019-10-28 10:03:18 -07:00
zhijxu
ce23d628a5 fix bug in cmake/onnxruntime_server.cmake 2019-10-28 10:03:18 -07:00
zhijxu
be7c24247f register execution provider when onnxruntime server creating sessions 2019-10-28 10:03:18 -07:00
pulkittomar
1fa956fb3f Undo integration test skip (#1917) 2019-10-27 09:47:31 -07:00
Yuri
a2596b706b FreeBSD compatibility patch. 
* Treat the 'amd64' architecture the same way as 'x86_64'
* Use thr_self() instead of gettid() on FreeBSD
 2019-10-26 12:44:12 -07:00
Dmitri Smirnov
88c58c19d4
Improve code readability and performance. (#2257) 
Improve code readability and performance. (#2257)  
  Remove one time checks from loops.
  Move out GetType<>() calls from loop as they
  go through local function statics.
  Get rid of index calculations from input and output
  so we can simlpy advance ptrs and potentially do better pre-fetch.
  Improve code readability.
 2019-10-25 16:19:59 -07:00
Changming Sun
ce14b07b1c
Fix the GPU nuget pipeline failure (#2255) 2019-10-25 13:55:38 -07:00
Sreekanth Yalachigere
c7599d0705 subgraph parse error fix (#2254) 2019-10-24 22:14:30 -07:00
shahasad
6a0ee7eff6
Fix model path marshalling in csharp, and re-enable the pretrained model tests (#2236) 2019-10-24 20:39:16 -07:00
edgchen1
8be48f47dd
Added Graph::ReplaceInitializedTensor() function. (#2230) 
Added Graph::ReplaceInitializedTensor() function to support updating existing initializers.
 2019-10-24 19:30:24 -07:00
Ryan Hill
7b4e6636f8 Add default params to Env C++ constructor (#2246) 2019-10-24 17:20:11 -07:00
Tomasz Dołbniak
63acd4e89b Adjust the nGraph EP to the newest CI test data (#2180) 
* Adjust the nGraph EP to the newest CI test data

* Increase the linux pipeline timeout for nGraph
 2019-10-24 16:44:03 -07:00
Ashwini Khade
e250e51e65
clear cache when moving to next channel (#2253) 2019-10-24 16:07:45 -07:00
Ryan Hill
7494500221 Fix csharp CXX sample (#2251) 2019-10-24 15:47:51 -07:00
suryasidd
f7b4bc15e1 Updated documentation for VAD-F (#2248) 
Signed-off-by: suryasidd <surya.siddharth.pemmaraju@intel.com>
 2019-10-24 14:31:44 -07:00
Scott McKay
20e6a2b6da
Disable optimizers for OpTester operator unit tests (#2237) 
* Disable optimizers for operator unit tests as they're intended to test the operator directly rather than something that could have been modified by an optimizer.

Disable TensorRT for Scan9 unit tests that fails when optimizers are enabled. Bug 525222 tracks that.

* Disable TRT for the lenient shape inferencing test as it uses Unsqueeze and TRT doesn't cope with that op.
 2019-10-24 11:37:09 -07:00
Ryan Hill
77d8d6f767
Remove the OrtApiBase base_ member from OrtApi (#2242) 
* Remove the OrtApiBase base_ member from OrtApi

* Forgot about C#
 2019-10-24 11:36:23 -07:00
edgchen1
6a27cb5ad6 Fixed tensor reference to const data and cleaned up Env API. (#1979) 2019-10-24 10:28:13 -07:00
Scott McKay
557243dce9
Add handling of zero dim to broadcasting and a few other ops (#2217) 
Handle dim value of zero in broadcasting, reduce, pool, conv, onehot, clip, *softmax, transpose ops on CPU and GPU. Add unit tests for all changes.
 2019-10-24 10:17:14 -07:00
Dmitri Smirnov
fef721c4f2
Update build instructions. Make samples build and run. (#2244) 
Making samples build. Update build instructions.
 2019-10-24 10:06:47 -07:00
kile0
bede664af7 mimalloc allocator (#2071) 2019-10-23 22:34:00 -07:00
Tracy Sharpe
90858b732a handle Concat-11 (#2240) 2019-10-23 19:12:40 -07:00
Pranav Sharma
eb75388cd5
Minor optimization: don't log all node names if they've all been assigned to a single provider. (#2229) 2019-10-23 15:16:15 -07:00
Scott McKay
41d55ea274
Update the GraphProto for subgraphs when saving the Graph. (#647) 
* Update the GraphProto for subgraphs when saving the Graph. This is required to produce a valid overall Graph if the Graph has been optimized.
 2019-10-23 15:14:06 -07:00