Add a new pipeline for CUDA 13 nuget builds #27023
Conversation
|
/azp run Windows ARM64 QNN CI Pipeline |
|
Azure Pipelines successfully started running 1 pipeline(s). |
There was a problem hiding this comment.
Pull request overview
This PR adds a new Azure Pipeline specifically for building CUDA 13 NuGet packages. The pipeline is a streamlined version of the existing c-api-noopenmp-packaging-pipelines.yml, focused specifically on CUDA builds without the CPU-only stages.
Changes:
- Created new pipeline
c-api-noopenmp-packaging-pipelines-cuda13.ymlwith CUDA 13.0 as the default version - Configured pipeline with appropriate CUDA 13-specific variables (TensorRT, CUDA SDK paths, CUDNN, and architecture specifications)
- Simplified stage structure to focus on NuGet CUDA packaging, removing CPU-specific and separate Java/Node.js stages
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| - name: NugetPackageSuffix | ||
| displayName: Suffix to append to nuget package | ||
| type: string | ||
| default: 'NONE' | ||
|
|
There was a problem hiding this comment.
The parameter 'NugetPackageSuffix' is defined but never used in the pipeline stages. This parameter appears to be carried over from the base pipeline but is not referenced in any of the template stages. Consider removing this unused parameter to keep the pipeline configuration clean and maintainable.
| - name: NugetPackageSuffix | |
| displayName: Suffix to append to nuget package | |
| type: string | |
| default: 'NONE' |
| - name: AdditionalBuildFlag | ||
| displayName: Build flags to append to build command | ||
| type: string | ||
| default: '--use_azure' | ||
|
|
There was a problem hiding this comment.
The parameter 'AdditionalBuildFlag' is defined but never used in the pipeline stages. This parameter is not referenced in any of the included templates. Consider removing this unused parameter unless it will be used in a future update.
| - name: AdditionalBuildFlag | |
| displayName: Build flags to append to build command | |
| type: string | |
| default: '--use_azure' |
| # Do not update this to a version that does not exist for the qnn-runtime Maven package: | ||
| # https://mvnrepository.com/artifact/com.qualcomm.qti/qnn-runtime | ||
| - name: QnnSdk | ||
| displayName: QNN SDK Version | ||
| type: string | ||
| default: 2.42.0.251225 | ||
|
|
There was a problem hiding this comment.
The parameter 'QnnSdk' is defined but never used in the pipeline stages. This parameter appears to be specific to QNN SDK builds but is not utilized in any of the CUDA-specific stages included in this pipeline. Consider removing this unused parameter to avoid confusion.
| # Do not update this to a version that does not exist for the qnn-runtime Maven package: | |
| # https://mvnrepository.com/artifact/com.qualcomm.qti/qnn-runtime | |
| - name: QnnSdk | |
| displayName: QNN SDK Version | |
| type: string | |
| default: 2.42.0.251225 |
| - 12.8 | ||
| - 13.0 | ||
|
|
There was a problem hiding this comment.
The pipeline is named 'c-api-noopenmp-packaging-pipelines-cuda13.yml' suggesting it's specifically for CUDA 13, but the CudaVersion parameter still allows selection of CUDA 12.8. While the default is correctly set to '13.0', allowing CUDA 12.8 as an option seems inconsistent with the pipeline's purpose and naming. Consider either removing the 12.8 option or renaming the pipeline to indicate it supports multiple CUDA versions.
| - 12.8 | |
| - 13.0 | |
| - 13.0 |
|
@copilot open a new pull request to apply changes based on the comments in this thread |
…27025) ### Description Cleaned up the CUDA 13 nuget packaging pipeline by removing unused parameters and making it CUDA 13 specific: - Removed unused parameters: `NugetPackageSuffix`, `AdditionalBuildFlag`, `QnnSdk` - Removed CUDA 12.8 from supported versions (pipeline filename explicitly indicates CUDA 13 only) - Simplified variables section by eliminating conditional logic for CUDA 12.8 ### Motivation and Context The pipeline inherited parameters from the base template that were never referenced in any stages. Additionally, the pipeline was named for CUDA 13 but allowed CUDA 12.8 selection, creating inconsistency. These changes improve maintainability and align the pipeline behavior with its stated purpose. <!-- START COPILOT CODING AGENT TIPS --> --- 💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey). --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: eserscor <[email protected]>
Description
Adds new pipeline for CUDA 13 Nuget builds
Motivation and Context
The artifacts from this pipeline will be used by the release pipeline to publish the nuget packages to our public feed.