-
Notifications
You must be signed in to change notification settings - Fork 3
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
Create a Debian package for Montblanc #250
Comments
The conda devs seem to have worked on somewhat similar problems: |
Yes, thanks for making this I priority. I would like to add that although it might not be possible at present time, it might be in the future. For that reason, I recommend keeping Montblanc compatible with Cuda 9.1 for now. But I know that also depends on what direction Tensorflow is going, which itself is a fast moving target. |
One option might be to create a Montblanc package with a dependency on the CPU version of tensorflow, eliminating the CUDA and cuDNN dependencies. |
would that still be useful? |
related issue here kernsuite/packaging#129 |
It would be useful to have a debian kernsuite package for Montblanc.
In future, Montblanc will have a hard dependency on tensorflow 1.6 (and perhaps 1.7, as a release candidate has just come out). This in turn creates dependencies on
CUDA 9.1 will be available in Ubuntu 18.04 bionic as a package. cuDNN 7.0 is a bit more tricky as I believe it is not redistributable (although debian packages are available here).
Part of the reason that I'm creating this issue is so that we can keep a record of conversations regarding packaging. I think the cuDNN dependency has come up in previous discussions and it would be good to remember these points.
The text was updated successfully, but these errors were encountered: