Skip to content

PyTorch, TensorFlow, Theano: Use a cuDNN version that has support for the CUDA version in fosscuda/2019a (CUDA 10.1)#9112

Merged
boegel merged 6 commits intoeasybuilders:developfrom
akesandgren:pytorch_fix_cudnn_version_for_2019a_toolchain
Oct 14, 2019
Merged

PyTorch, TensorFlow, Theano: Use a cuDNN version that has support for the CUDA version in fosscuda/2019a (CUDA 10.1)#9112
boegel merged 6 commits intoeasybuilders:developfrom
akesandgren:pytorch_fix_cudnn_version_for_2019a_toolchain

Conversation

@akesandgren
Copy link
Copy Markdown
Contributor

@akesandgren akesandgren commented Oct 11, 2019 •
edited by boegel
Loading

Also fixes incorrect postinstallcmds for both foss and fosscuda versions.

requires #9108 (cuDNN)

@akesandgren akesandgren added this to the 4.x milestone Oct 11, 2019
@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 11, 2019

Test report by @akesandgren
FAILED
Build succeeded for 0 out of 1 (1 easyconfigs in this PR)
b-an03.hpc2n.umu.se - Linux ubuntu 16.04, Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz, Python 2.7.12
See https://gist.github.com/1dd6c848632daed7192be6b37c6f966e for a full test report.

@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 12, 2019

The cuDNN version problem will of course go away once all 4 of the cuDNN version bugfixes are merged.

@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 12, 2019

Test report by @akesandgren
SUCCESS
Build succeeded for 2 out of 2 (2 easyconfigs in this PR)
b-an03.hpc2n.umu.se - Linux ubuntu 16.04, Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz, Python 2.7.12
See https://gist.github.com/cff6bb22978b1f40ac7224a50cfb3ecd for a full test report.

@easybuilders easybuilders deleted a comment from boegelbot Oct 12, 2019
@boegel
Copy link
Copy Markdown
Member

boegel commented Oct 12, 2019

@akesandgren This is difficult to merge because of the broken tests, why not collapse all 4 PRs into one (here for example), so the tests pass?

That makes it a bit more painful w.r.t. submitting a test report, but it's a lot easier to get it merged...

@boegel boegel modified the milestones: 4.x, 4.0.1 Oct 12, 2019
@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 12, 2019

Merged the fix for PR #9113, #9114, and #9115 into this one to make merging into develop possible, avoiding cuDNN version problems.

This was referenced Oct 12, 2019
Closed
Closed
Closed
@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 12, 2019

Test report in progress

@easybuilders easybuilders deleted a comment from boegelbot Oct 12, 2019
@akesandgren akesandgren changed the title PyTorch: Use a cuDNN version that has support for the CUDA version in… PyTorch, TensorFlow, Theano: Use a cuDNN version that has support for the CUDA version in… Oct 12, 2019
@easybuilders easybuilders deleted a comment from boegelbot Oct 12, 2019
@akesandgren
Copy link
Copy Markdown
Contributor Author

akesandgren commented Oct 12, 2019

Test report by @akesandgren
SUCCESS
Build succeeded for 5 out of 5 (5 easyconfigs in this PR)
b-an03.hpc2n.umu.se - Linux ubuntu 16.04, Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz, Python 2.7.12
See https://gist.github.com/7c54a7edb467212187da8064ea8afed4 for a full test report.

@zao
Copy link
Copy Markdown
Contributor

zao commented Oct 12, 2019

Test report by @zao
SUCCESS
Build succeeded for 7 out of 7 (5 easyconfigs in this PR)
freja - Linux ubuntu 18.04, Intel(R) Core(TM) i7-8700K CPU @ 3.70GHz, Python 2.7.15+
See https://gist.github.com/23d87d527d854fbc390ba74d793a8bef for a full test report.

@boegel boegel changed the title PyTorch, TensorFlow, Theano: Use a cuDNN version that has support for the CUDA version in… PyTorch, TensorFlow, Theano: Use a cuDNN version that has support for the CUDA version in fosscuda/2019a (CUDA 10.1) Oct 13, 2019
boegel
boegel approved these changes Oct 13, 2019
View reviewed changes
Copy link
Copy Markdown
Member

@boegel boegel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

@boegel
Copy link
Copy Markdown
Member

boegel commented Oct 13, 2019

Test report by @boegel
SUCCESS
Build succeeded for 5 out of 5 (5 easyconfigs in this PR)
node3301.joltik.os - Linux centos linux 7.6.1810, Intel(R) Xeon(R) Gold 6242 CPU @ 2.80GHz, Python 2.7.5
See https://gist.github.com/dbce7a6f1e7403d285321a63b681d56a for a full test report.

@boegel
Copy link
Copy Markdown
Member

boegel commented Oct 14, 2019

Going in, thanks @akesandgren!

@boegel boegel merged commit f7b2972 into easybuilders:develop Oct 14, 2019
@akesandgren akesandgren deleted the pytorch_fix_cudnn_version_for_2019a_toolchain branch October 14, 2019 06:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@boegel boegel boegel approved these changes

Assignees

No one assigned

Labels

bug fix

Projects

None yet

Milestone

4.0.1

Development

Successfully merging this pull request may close these issues.

3 participants

@akesandgren @boegel @zao