Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions .github/workflows/test-software.eessi.io.yml
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,7 @@ jobs:
- aarch64/neoverse_v1
- x86_64/amd/zen2
- x86_64/amd/zen3
- x86_64/amd/zen4
- x86_64/intel/haswell
- x86_64/intel/skylake_avx512
- x86_64/generic
Expand Down Expand Up @@ -58,6 +59,15 @@ jobs:
# first check the CPU-only builds for this CPU target
echo "just run check_missing_installations.sh (should use easystacks/software.eessi.io/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-*.yml)"
for easystack_file in $(ls easystacks/software.eessi.io/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-eb-*.yml); do
if [ ${{matrix.EESSI_SOFTWARE_SUBDIR_OVERRIDE}} = "x86_64/amd/zen4" ]; then
if grep -q 2022b <<<"${easystack_file}"; then
Copy link
Contributor

Choose a reason for hiding this comment

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

@laraPPr Makes sense, but let's add a comment to explain why we're doing this, both for 2022b and CUDA?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

added some comments

Copy link
Collaborator

Choose a reason for hiding this comment

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

I'm still confused, but that might be me. We're also not building CUDA stuff for e.g. haswell right now, but we don't have an explicit skip there, right? Why do we need it for zen4?

Copy link
Contributor

Choose a reason for hiding this comment

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

We have the CUDA installations of CUDA-Samples and OSU-MicroBenchmarks installed for all CPU targets in the CPU-only path, but we don't want to do that going forward.

Those installations were split off into separate easystack files in #762 to make this possible.

Once we've made some more progress with the accel/nvidia part of the software layer, we'll likely remove those CUDA installations from the CPU path, they don't belong there.

# skip the check of installed software on zen4 for foss/2022b builds
continue
elif grep -q CUDA <<<"${easystack_file}"; then
# skip the check of install CUDA software in the CPU path for zen4
continue
fi
fi
echo "check missing installations for ${easystack_file}..."
./check_missing_installations.sh ${easystack_file}
ec=$?
Expand Down