TMP 2022-10-13 Meeting notes

Andrew

New docker container doing well, 3 tests are failing that don’t fail elsewhere. One involves a mutex and worker thread reuse, an old bug with test code reproduces, the old bug was never fixed. Ran on a thunderx2 not in the docker container and hasn’t failed yet. Number of cores might be involved as the docker case is on a 200+ core machine but the thunderx2 and intel haven't failed and have far fewer cores.

Crefeda

Been out a bit, getting back to speed.

Tom

Performance benchmarks are looking good, improves from latest measures in a few places, +4.5% on the high end and biggest regression ~2%. See : airspeed velocity