gemini-code-assist[bot] commented on code in PR #18783:
URL: https://github.com/apache/tvm/pull/18783#discussion_r2809548173


##########
docker/Dockerfile.ci_gpu:
##########
@@ -17,7 +17,7 @@
 
 # CI docker GPU env
 # tag: v0.60
-FROM nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04
+FROM nvidia/cuda:12.8.1-cudnn-devel-ubuntu22.04

Review Comment:
   ![high](https://www.gstatic.com/codereviewagent/high-priority.svg)
   
   While updating the CUDA version to 12.8.1 is the correct fix for the 
described cuDNN issue, this change may introduce inconsistencies with other 
dependencies that are installed in this Dockerfile. These dependencies might be 
pinned to older CUDA versions.
   
   For example:
   - **JAX**: The installation script `docker/install/ubuntu_install_jax.sh` 
seems to install JAX for CUDA 11 (`jax[cuda11_pip]`). This should be updated to 
a version compatible with CUDA 12.x (e.g., `jax[cuda12_pip]`).
   - **PyTorch**: The script `docker/install/ubuntu_install_onnx.sh` installs a 
version of PyTorch that was built for older CUDA versions (11.8/12.1). This 
might not be fully compatible with CUDA 12.8.1.
   
   To ensure the stability of the CI environment and prevent potential runtime 
errors in other tests, I recommend updating these dependencies to align with 
the new CUDA version in this same pull request.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to