[NV RTX EP] Upstream changes from the win-ort #25370
Conversation
There was a problem hiding this comment.
Pull Request Overview
This PR replaces references to win-onnxruntime with onnxruntime, restores build integrity, and adds auto EP selection support for the TensorRT RTX provider.
- Added a helper and test (
AutoEp_PreferGpu) to verify automatic GPU EP selection. - Exported new symbols (
CreateEpFactories,ReleaseEpFactory) and implemented OrtEpFactory support. - Updated the
CUDAPinnedAllocatorconstructor argument order to match the new signature.
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| onnxruntime/test/providers/nv_tensorrt_rtx/nv_basic_test.cc | Added SessionHasEp helper and AutoEp_PreferGpu test |
| onnxruntime/core/providers/nv_tensorrt_rtx/symbols.def | Exported CreateEpFactories and ReleaseEpFactory symbols |
| onnxruntime/core/providers/nv_tensorrt_rtx/nv_provider_factory.cc | Implemented CreateIExecutionProvider override and OrtEpFactory functions |
| onnxruntime/core/providers/nv_tensorrt_rtx/nv_execution_provider.cc | Updated argument order for CUDAPinnedAllocator construction |
Comments suppressed due to low confidence (3)
onnxruntime/test/providers/nv_tensorrt_rtx/nv_basic_test.cc:409
- This test only works on Windows but isn’t guarded; wrap it in a
#ifdef _WIN32to avoid failures on other platforms.
TEST(NvExecutionProviderTest, AutoEp_PreferGpu) {
onnxruntime/core/providers/nv_tensorrt_rtx/nv_provider_factory.cc:238
- Public factory functions should use the
ORT_APIexport macro (asGetProviderdoes) to ensure consistent symbol visibility across platforms.
OrtStatus* CreateEpFactories(const char* /*registration_name*/, const OrtApiBase* ort_api_base,
onnxruntime/core/providers/nv_tensorrt_rtx/nv_provider_factory.cc:258
- Likewise,
ReleaseEpFactoryshould be annotated withORT_APIto match the export conventions and prevent missing-symbol issues.
OrtStatus* ReleaseEpFactory(OrtEpFactory* factory) {
jywu-msft
changed the title
|
/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows x64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 5 pipeline(s). |
|
Merged to latest main to fix build failures. |
|
/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline, Windows x64 QNN CI Pipeline |
|
Azure Pipelines successfully started running 5 pipeline(s). |
|
/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline, Windows x64 QNN CI Pipeline |
|
Azure Pipelines successfully started running 5 pipeline(s). |
|
/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline, Windows x64 QNN CI Pipeline |
|
Azure Pipelines successfully started running 5 pipeline(s). |
|
pls fix lint error |
|
/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline, Windows x64 QNN CI Pipeline |
|
Azure Pipelines successfully started running 5 pipeline(s). |
|
Hi there! We haven't cut the release branch for this version yet, so I'm removing the |
### Description <!-- Describe your changes. --> Changes from win-onnxruntime to onnxruntime Fix the build break. --------- Co-authored-by: Yulong Wang <[email protected]>
### Description <!-- Describe your changes. --> Changes from win-onnxruntime to onnxruntime Fix the build break. --------- Co-authored-by: Yulong Wang <[email protected]>
Description
Changes from win-onnxruntime to onnxruntime
Fix the build break.