Frontier Benchmarking (#453) #881
Conversation
.github/workflows/bench.yml
Outdated
|
|
||
| frontier: | ||
| name: Oak Ridge | Frontier (AMD ROCm) | ||
| if: github.repository == 'MFlowCode/MFC' && needs.file-changes.outputs.checkall == 'true' |
There was a problem hiding this comment.
this almost entirely duplicates the code from phoenix... can you make a CI matrix instead to avoid the extra lines? also there is no need to measure performance on CPU cases on Frontier, that's just going to slow down the number of cases queued. i only have 1 frontier runner because that's the max num. of debug jobs one can submit.
There was a problem hiding this comment.
Yeah sure, I will apply those changes shortly.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #881 +/- ##
=======================================
Coverage 45.76% 45.76%
=======================================
Files 68 68
Lines 18668 18668
Branches 2251 2251
=======================================
Hits 8543 8543
Misses 8767 8767
Partials 1358 1358 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
.github/workflows/bench.yml
Outdated
| device: gpu | ||
| build_script: "" | ||
| - cluster: frontier | ||
| name: Oak Ridge | Frontier (AMD CCE) |
|
Reduced the job duration to 3 hrs to see whether it would yield the same error regardless of duration. |
Description
Added two benchmarking cases by submitting SLURM jobs on Frontier - duplicate implementation of Phoenix. (#453)