CUDA: fix broken oob check for FA vec f32 kernel (#7904) #25
| Job | Run time |
|---|---|
| 2m 1s | |
| 2m 47s | |
| 1m 27s | |
| 1m 28s | |
| 2m 8s | |
| 1m 38s | |
| 9m 4s | |
| 11m 3s | |
| 14m 12s | |
| 44m 54s | |
| 11m 36s | |
| 11m 39s | |
| 2m 5s | |
| 2m 55s | |
| 1m 25s | |
| 2m 38s | |
| 2m 27s | |
| 3m 46s | |
| 5m 8s | |
| 8m 35s | |
| 5m 56s | |
| 3m 45s | |
| 6m 11s | |
| 18m 7s | |
| 1m 6s | |
| 7m 7s | |
| 22m 24s | |
| 7m 13s | |
| 14m 10s | |
| 6m 2s | |
| 31m 15s | |
| 12m 40s | |
| 7m 50s | |
| 8m 30s | |
| 17m 30s | |
| 7m 52s | |
| 6m 31s | |
| 7m 44s | |
| 6m 16s | |
| 4m 49s | |
| 2m 39s | |
| 1m 0s | |
| 5h 49m 33s |