Repository navigation
Conversation
EESSI/2023.06 uses the older cuDDN version and is now failing the check, even already at the installation of CUDA libraries to `host_injections`.
casparvl
approved these changes
Oct 14, 2026
casparvl
left a comment
casparvl
left a comment
Contributor
There was a problem hiding this comment.
I already replicated the issue here https://gitlab.com/eessi/support/-/work_items/283#note_4004603381
To test the fix, I checked out your feature branch, and made a small hard-coded modification to the install_cuda_and_libraries.sh from that branch to overwrite the EASYBUILD_HOOKS by the one from this feature branch:
EESSI_SITE_CPU_FAMILY_PATH=$(echo "$EESSI_SITE_SOFTWARE_PATH" | sed 's|\(.*\)'"$EESSI_VERSION"/software/"$EESSI_OS_TYPE"/"$EESSI_SOFTWARE_SUBDIR"'|\1'"$EESSI_CPU_FAMILY"'|')
export EASYBUILD_INSTALLPATH=$EESSI_SITE_CPU_FAMILY_PATH
# FOR DEBUGGING ONLY!
export EASYBUILD_HOOKS=/scratch-local/casparl.28074581/software-layer-scripts/eb_hooks.py
# Install modules in hidden .modules dir to keep track of what was installed before
# (this action is temporary, and we do not call Lmod again within the current shell context, but in EasyBuild
# subshells, so loaded modules are not automatically unloaded)
MODULEPATH=${EASYBUILD_INSTALLPATH}/.modules/all
echo "set MODULEPATH=${MODULEPATH}"
I then ran this from the local checkout using:
NVIDIA_EASYSTACKS_DIRECTORY=scripts/gpu_support/nvidia/easystacks/2023.06/ scripts/gpu_support/nvidia/install_cuda_and_libraries.sh --accept-cuda-eula --accept-cudnn-eula
In the config output, I see:
cuda-compute-capabilities (E) = 9.0a
cuda-sanity-check-error-on-failed-checks (E) = True
as expected. Then, I see:
== 2026-10-14 15:48:22,471 config.py:712 WARNING Build option 'cuda_compute_capabilities' was updated to: ['9.0']
In the cuDNN build log, showing this works as intended. Indeed, build now passes:
== Build succeeded for 1 out of 1 (total: 1 min 59 secs)
== Summary:
* [SUCCESS] cuDNN/8.9.2.26-CUDA-12.1.1
== Temporary log file(s) /scratch-local/casparl.28081376/eb-oztndv2n/easybuild-w79_2z_o.log* have been removed.
== Temporary directory /scratch-local/casparl.28081376/eb-oztndv2n has been removed.
all installations at /cvmfs/software.eessi.io/host_injections/x86_64/software/... succeeded!
Contributor
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
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
2 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.
EESSI/2023.06 uses the older cuDDN version and is now failing the check, even already at the installation of CUDA libraries to
host_injections.