Installer Update with Cuda 12, Latest Trt support #285
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The key changes made in the updated installation script~: 1. **Refactoring and Simplification**: The updated script has been refactored for better readability and maintainability. Functions like `get_installed_version` and `install_package` are introduced to reduce code repetition and make the script more modular. 2. **Version Checks and Updates**: - The `tensorrt` package version has been updated from `9.0.1.post11.dev4` to `9.3.0.post12.dev1`. This change indicates a significant version update, possibly including new features, bug fixes, and performance improvements. - The CUDA Deep Neural Network library (`nvidia-cudnn-cu11`) dependency has been replaced with `nvidia-cudnn-cu12` in the updated script, suggesting a move to support newer CUDA versions (`cu12` instead of `cu11`). However, there seems to be a discrepancy in the version handling (`8.9.7.29` vs. `8.9.6.50`) that might be a typo or an intentional downgrade for compatibility reasons. 3. **Improved Package Management**: - The updated script includes more sophisticated logic for managing package installations, including handling uninstallation of previous versions if needed, before installing a new version. This is especially evident in the handling of the `tensorrt` and `nvidia-cudnn-cu12` packages. - The addition of `no_cache_dir=True` to all `install_package` calls to ensure the latest versions are fetched and to avoid potential conflicts with cached packages. 4. **Simplified Dependency Handling**: - The dependencies for ONNX Graph Surgeon and Polygraphy are handled more straightforwardly in the updated script, with specific versions for dependent packages like `protobuf`. 5. **Error Handling and Robustness**: The introduction of the `get_installed_version` function allows for more robust error handling by checking if a package is installed and obtaining its version before attempting to install or uninstall. This reduces the risk of errors during the installation process.
ranareehanaslam
commented
Mar 20, 2024
Author
ranareehanaslam
left a comment
ranareehanaslam
left a comment
There was a problem hiding this comment.
nvidia-cudnn-cu12 check version match 8.9.6.50
ice-fly
reviewed
May 10, 2025
There was a problem hiding this comment.
Might reccomend reconsidering the get-intsalled-version function defaults (no_cache_dir gets passed to the function with true rather than being default) Otherwise good code.
ice-fly
reviewed
May 10, 2025
ice-fly
approved these changes
May 10, 2025
ice-fly
left a comment
ice-fly
left a comment
There was a problem hiding this comment.
Othewise excellent pull request, reccomend approval
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The key changes made in the updated installation script~:
Refactoring and Simplification: The updated script has been refactored for better readability and maintainability. Functions like
get_installed_versionandinstall_packageare introduced to reduce code repetition and make the script more modular.Version Checks and Updates:
tensorrtpackage version has been updated from9.0.1.post11.dev4to9.3.0.post12.dev1. This change indicates a significant version update, possibly including new features, bug fixes, and performance improvements.nvidia-cudnn-cu11) dependency has been replaced withnvidia-cudnn-cu12in the updated script, suggesting a move to support newer CUDA versions (cu12instead ofcu11). However, there seems to be a discrepancy in the version handling (8.9.7.29vs.8.9.6.50) that might be a typo or an intentional downgrade for compatibility reasons.Improved Package Management:
tensorrtandnvidia-cudnn-cu12packages.no_cache_dir=Trueto allinstall_packagecalls to ensure the latest versions are fetched and to avoid potential conflicts with cached packages.Simplified Dependency Handling:
protobuf.Error Handling and Robustness: The introduction of the
get_installed_versionfunction allows for more robust error handling by checking if a package is installed and obtaining its version before attempting to install or uninstall. This reduces the risk of errors during the installation process.