TensorFlow Docker image uses cuDNN 9, but only cuDNN 8.9 is installed
Nobody has claimed this yet.
- Dominant language
- Shell
- Stars
- 311
- Forks
- 139
- Avg merge
- 1h 28m
- Merged PRs (30d)
- 1
Description
Starts from tensorflow/tensorflow-2.15.0-gpu (206b54412f00), tensorflow tries to load libcudnn.so.9, but the docker image uses 8:
https://github.com/tensorflow/build/blob/master/tensorflow_runtime_dockerfiles/gpu.packages.txt#L15-L17
For more information, please see:
https://github.com/tensorflow/tensorflow/issues/62412
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with tensorflow_runtime_dockerfiles/gpu.packages.txt at the linked lines and inspect how the tensorflow/tensorflow-2.15.0-gpu image installs its cuDNN packages. Reproduce or inspect the image to confirm the library-version mismatch, then verify that the image's installed cuDNN version matches the version TensorFlow attempts to load.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- docker, tensorflow
- Domain
- build-system, devops
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Clearly specified
- Newbie friendliness
- 45/100