[Pallas/Mosaic GPU] Implement a more comprehensive matmul kernel to see what we're still missing #62475
Job | Run time |
---|---|
3m 51s | |
1m 24s | |
42s | |
3m 48s | |
1m 12s | |
55s | |
2m 30s | |
14m 22s |
Job | Run time |
---|---|
3m 51s | |
1m 24s | |
42s | |
3m 48s | |
1m 12s | |
55s | |
2m 30s | |
14m 22s |