Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Cellpose update: 3.0.1 --> 3.1.0 #587

Merged
merged 1 commit into from
Nov 22, 2024

Conversation

kbestak
Copy link
Contributor

@kbestak kbestak commented Nov 3, 2024

Submitting a Container

(If you're requesting for a new container, please check the procedure described here.

Check BioContainers' Dockerfile specifications

Checklist

  1. Misc
  • My tool doesn't exist in BioConda
  • The image can be built
  1. Metadata
  • LABEL base_image
  • LABEL version
  • LABEL software.version
  • LABEL about.summary
  • LABEL about.home
  • LABEL about.license
  • MAINTAINER
  1. Extra (optionals)
  • I have written tests in test-cmds.txt
  • LABEL extra.identifier
  • LABEL about.documentation
  • LABEL about.license_file
  • LABEL about.tags

Check BioContainers' Dockerfile metadata

Changes: https://github.com/MouseLand/cellpose/releases/tag/v3.1.0

@kbestak
Copy link
Contributor Author

kbestak commented Nov 3, 2024

Hi Biocontainers team,
I have two questions connected to the PR above:

  1. What would be the best way to add GPU acceleration support to the container? I was able to build an image from an nvidia base image instead of python (as is in the Dockerfile above) that can utilize available GPUs, however the size is then much larger.
  2. Is it acceptable to have a CPU-only and GPU option for the same version of a tool?
    Thank you!

@mboudet
Copy link
Contributor

mboudet commented Nov 7, 2024

Hi,

I believe the issue withe GPU support is that you need to match the CUDA version in the container to the host GPU driver, so it's difficult to build in a docker (you would need to build a container for each major cuda version).

Not sure if they now manage backward / forward version support.

As for building separate version, if there are no other choices (ie, the tool built in GPU mode cannot support CPU mode), I'm not against it.

@kbestak
Copy link
Contributor Author

kbestak commented Nov 8, 2024

From testing, the container with CUDA can still run on CPU if GPU is not provided, however since the container is then much larger, I thought it would be good to have both. I can build both and come back with the estimated container size.

@mboudet
Copy link
Contributor

mboudet commented Nov 8, 2024

Ok. There is still the issue of the CUDA compatibility. I'd rather not have to build a container for each CUDA main version..

@sminot
Copy link
Contributor

sminot commented Nov 21, 2024

I would love to be able to use cellulose 3.1.0, if there is any way to merge this PR without waiting on a solution for CUDA

@kbestak
Copy link
Contributor Author

kbestak commented Nov 22, 2024

From my side, I'd suggest setting the CUDA topic aside in that case.

@mboudet
Copy link
Contributor

mboudet commented Nov 22, 2024

That's fine by me

@biocontainers-bot
Copy link
Collaborator

No test-cmds.txt (test file) present, skipping tests

@mboudet
Copy link
Contributor

mboudet commented Nov 22, 2024

Alright, should be fine to merge. Thanks for your contribution.

@mboudet mboudet merged commit c470731 into BioContainers:master Nov 22, 2024
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants