add fused transpose and non-transpose kernel and use it for grad outp… #1125
| Job | Run time |
|---|---|
| 8s | |
| 3m 10s | |
| 10m 43s | |
| 8m 44s | |
| 8m 57s | |
| 16m 15s | |
| 19m 18s | |
| 5m 53s | |
| 2m 59s | |
| 11m 0s | |
| 4m 18s | |
| 8m 54s | |
| 15m 53s | |
| 18m 57s | |
| 5m 53s | |
| 2m 57s | |
| 10m 58s | |
| 9m 59s | |
| 9m 4s | |
| 15m 58s | |
| 18m 58s | |
| 5m 54s | |
| 2m 59s | |
| 10m 46s | |
| 4m 24s | |
| 8m 54s | |
| 15m 57s | |
| 18m 41s | |
| 5m 54s | |
| 2m 58s | |
| 10m 55s | |
| 9m 12s | |
| 8m 36s | |
| 16m 4s | |
| 18m 45s | |
| 5m 51s | |
| 20s | |
| 14s | |
| 17s | |
| 15s | |
| 21s | |
| 15s | |
| 19s | |
| 16s | |
| 15s | |
| 21s | |
| 31s | |
| 19s | |
| 14s | |
| 15s | |
| 17s | |
| 25s | |
| 18s | |
| 26s | |
| 19s | |
| 15s | |
| 14s | |
| 23s | |
| 18s | |
| 14s | |
| 18s | |
| 14s | |
| 18s | |
| 19s | |
| 16s | |
| 19s | |
| 19s | |
| 18s | |
| 20s | |
| 19s | |
| 16s | |
| 6h 5m 23s |