gswangg
gswangg
Update -- been short on time due to travel, but making progress. I've got the issue repro'ing locally with gpuocelot in a docker. The particular assertion we're seeing in CI...
I'm working on a WMMA-related process_replay regression introduced after merging upstream master in my dev branch. Should have it figured out tomorrow.
@Qazalin in light of geohot's comment in #5729, I reverted my process replay hacks in the PTX renderer and kept the change needed to functionally pass the tests. I can...
Woohoo checks passed! Should I mark this PR ready for review?