saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-29 20:14:01 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Update TensorRT parser (#3650)

Browse source 
* update onnx-tensorrt submodule

* add more model dumping point

* update trt kernel name and docker readme file

* fix minior issues

* fix format issue

* update onnx-tensorrt submodule

Co-authored-by: stevenlix <stevenlix>
This commit is contained in:
stevenlix 2020-04-23 20:51:44 -07:00 committed by GitHub
parent 939d036660
commit 2ab78c5da1
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 9 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
cmake/external/onnx-tensorrt vendored

@ -1 +1 @@
Subproject commit 8b8b89d18423cae7bb8a6a95ea346dd49a4dccc0
Subproject commit 177a2e470ca28433b8d8ca30e44f79b373f2407e

2
dockerfiles/README.md
View file

@ -80,7 +80,7 @@ Use `docker pull` with any of the images and tags below to pull an image and try
```
## TensorRT
**Ubuntu 18.04, CUDA 10.1.243, TensorRT 6.0.1**
**Ubuntu 18.04, CUDA 10.2, TensorRT 7.0.0**
1. Build the docker image from the Dockerfile in this repository.
```

8
onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc
View file

@ -335,7 +335,7 @@ std::unique_ptr<IndexedSubGraph> TensorrtExecutionProvider::GetSubGraph(SubGraph
// Assign inputs and outputs to subgraph's meta_def
auto meta_def = onnxruntime::make_unique<::onnxruntime::IndexedSubGraph::MetaDef>();
meta_def->name = "TRTKernel_" + std::to_string(kernels_index++);
meta_def->name = "TRTKernel_" + graph.Name() + "_" + std::to_string(kernels_index++);
meta_def->domain = kMSDomain;
for (const auto& input : inputs) {
@ -454,6 +454,12 @@ SubGraphCollection_t TensorrtExecutionProvider::GetSupportedList(SubGraphCollect
std::string string_buf;
model_proto.SerializeToString(&string_buf);
if (dump_subgraphs_) {
// Dump TensorRT subgraph for debugging if enabled via ORT_TENSORRT_DUMP_SUBGRAPHS env variable.
std::fstream dump("TensorrtExecutionProvider_TRT_Subgraph.onnx", std::ios::out | std::ios::trunc | std::ios::binary);
model_proto.SerializeToOstream(&dump);
}
// Get supported node list recursively
SubGraphCollection_t parser_nodes_list;
TensorrtLogger& trt_logger = GetTensorrtLogger();