enhance TensorFlow easyblock to support installing TF 1.14.0 with CUDA and MPI support#1811
Merged
akesandgren merged 3 commits intoeasybuilders:developfrom Sep 26, 2019
Conversation
…w easyblock to issue with installation of TensorFlow 1.14.0 with CUDA support
damianam
previously requested changes
Sep 24, 2019
| }) | ||
|
|
||
| # for recent TensorFlow versions, $TF_CUDA_PATHS and $TF_CUBLAS_VERSION must also be set | ||
| if LooseVersion(self.version) > LooseVersion('1.14'): |
Member
There was a problem hiding this comment.
This seems more intuitive for me:
if LooseVersion(self.version) >= LooseVersion('1.14.0'):
Other than that lgtm!
Member
Author
There was a problem hiding this comment.
@damianam Good point here actually, this should be >= to avoid confusion/surprises...
Code does the right thing as is. Dismissing so it can be merged.
Contributor
|
Going in, thanks @boegel! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This fixes the following issue when trying to install TensorFlow 1.14.0 with CUDA 10.1.243:
Changes were made in TensorFlow 1.14.0 that require both
$TF_CUDA_VERSIONand$TF_CUDNN_VERSIONto only include the correct number of digits, so10.1rather than10.1.243forCUDA.In older TensorFlow versions, the value specified in
$TF_CUDA_VERSIONwas stripped down first to just<major>.<minor>, but this got lost in TF 1.14.0.For
cuDNN, it expects<major>.<minor>.<patch>(so 3, not 4 as in thecuDNNversion we use).It turns out that we've been doing it incorrectly all along, since even with these changes installing
TensorFlow-1.8.0-fosscuda-2017b-Python-2.7.14.ebstill works fine...