feat(cuda): fuse narrower-than-output Dict codes and RunEnd ends (#7603) #5408
| Job | Run time |
|---|---|
| 37s | |
| 11m 49s | |
| 34m 7s | |
| 29m 5s | |
| 7m 31s | |
| 19m 10s | |
| 15m 22s | |
| 4m 21s | |
| 13m 9s | |
| 32m 26s | |
| 3m 36s | |
| 9m 27s | |
| 7m 8s | |
| 3h 7m 48s |
| Job | Run time |
|---|---|
| 37s | |
| 11m 49s | |
| 34m 7s | |
| 29m 5s | |
| 7m 31s | |
| 19m 10s | |
| 15m 22s | |
| 4m 21s | |
| 13m 9s | |
| 32m 26s | |
| 3m 36s | |
| 9m 27s | |
| 7m 8s | |
| 3h 7m 48s |