AMD support
Opened this issue · 2 comments
wsippel commented
I played around with AMD's AITemplate fork on my 7900XTX a while ago, and it seems to be the fastest Stable Diffusion implementation on AMD hardware right now. That said, RDNA3 support hasn't been upstreamed yet, but it is available here: navi3_rel_ver_1.0
wsippel commented
Managed to manually merge AMD's branch into upstream, but it segfaults in libamdhip64 if I try to compile any AIT modules. No idea why, it used to work a couple of weeks ago with the SD example bundled with AITemplate. Maybe a regression in Torch or ROCm.
FizzleDorf commented
This is a good find! Thank you so much for sharing! There are several issues to iron out before I can look into AMD support.