Performance test T9630 is failing
Recently T9630 has started failing on CI. Observing the latest CI results from the past 500 commits I get this graph tmpzux4zjjm.html. It's clear that the variance has largely increased since about 20 commits ago. You can generate this graph with this command (depend on !996 (closed)):
$ git fetch https://gitlab.haskell.org/ghc/ghc-performance-notes.git refs/notes/perf:refs/notes/ci/perf
$ python3 testsuite/driver/perf_notes.py --chart --ci --test-name T9630 --test-env i386-linux-deb9 master~500..master
You can also remove --test-env i386-linux-deb9
to see all test environments.
This change needs to be characterized. Is this a change in CI load, or is there really something going on with this performance test?