Remove the default-loaded module craype-hugepages2M which interferes with GNU on Cori
#3129
Conversation
This file contains 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
…loaded after Cori upgrade as it causes issues with GNU-built executables. In the past, the hugepages module had no impact on our performance. Also, remove the PERL env variable that was being temporarily used to point to our own copy of the Switch PERL module as it should now be installed on Cori's login nodes. [bfb]
…n Cori login nodes.
|
hugepages is also causing memory failures (where they did not occur before) in F cases with very large meshes, e.g. ne1024 . |
|
Are we sure we can point to the hugepages module as the reason for failure? I think Cray/NERSC would certainly not expect there to be any faults from this module -- doesn't mean it's not there. |
|
This solved my build failure, and it is the only change I made. |
ndkeen
added a commit
that referenced
this pull request
Aug 15, 2019
After Cori software upgrade, one NERSC change was to load the craype-hugepages2M and this PR would simply add a command to explicitly remove that module for cori-knl and cori-haswell. These are a few hugepages modules (at different sizes 2M,4M8M, etc) that can be loaded and require a rebuild to use. The use of hugepages for some apps can impact performance variability and reduce the cost of accessing memory in several MPI_Alltoall operations. However, initial tests when porting to Cori showed that there was no performance difference using these. I would just leave the module loaded (in favor of default machine settings), but for whatever reason, our GNU-built executables are failing unless the module is unloaded. [bfb] Fixes #3127
|
merged to next |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
After Cori software upgrade, one change was to load the
craype-hugepages2Mand this PR would simply add a command to explicitly remove that module for cori-knl and cori-haswell.These are a few hugepages modules (at different sizes 2M,4M8M, etc) that can be loaded and require a rebuild to use. The use of hugepages for some apps can impact performance variability and reduce the cost of accessing memory in several MPI_Alltoall operations. However, initial tests when porting to Cori showed that there was no performance difference using these. I would just leave the module loaded (in favor of default machine settings), but for whatever reason, our GNU-built executables are failing unless the module is unloaded.
[bfb]
Fixes #3127