CUDA: remove DMMV, consolidate F16 mult mat vec #289
| Job | Run time |
|---|---|
| 11m 48s | |
| 3m 54s | |
| 3m 23s | |
| 2m 32s | |
| 11m 16s | |
| 1m 38s | |
| 1m 43s | |
| 1m 51s | |
| 3m 46s | |
| 18m 8s | |
| 11m 40s | |
| 2m 58s | |
| 4m 41s | |
| 3m 16s | |
| 4m 31s | |
| 3m 18s | |
| 2m 32s | |
| 2m 28s | |
| 3m 42s | |
| 3m 7s | |
| 6m 41s | |
| 4m 51s | |
| 2m 32s | |
| 2m 50s | |
| 7m 22s | |
| 1m 28s | |
| 4m 41s | |
| 4m 27s | |
| 5m 10s | |
| 5m 56s | |
| 5m 58s | |
| 4m 41s | |
| 5m 32s | |
| 4m 53s | |
| 3m 16s | |
| 46m 26s | |
| 46m 16s | |
| 10m 33s | |
| 13m 29s | |
| 28m 48s | |
| 0s | |
| 9m 12s | |
| 0s | |
| 5h 27m 14s |