Support the case where query_len>context_len in the multi-queries paged attention. #10308
Job | Run time |
---|---|
3s | |
33m 25s | |
0s | |
1m 35s | |
8m 33s | |
13m 26s | |
4m 31s | |
8m 1s | |
10m 38s | |
5m 0s | |
5m 37s | |
1h 30m 49s |
Job | Run time |
---|---|
3s | |
33m 25s | |
0s | |
1m 35s | |
8m 33s | |
13m 26s | |
4m 31s | |
8m 1s | |
10m 38s | |
5m 0s | |
5m 37s | |
1h 30m 49s |