change argmax to DeviceSegmentedReduce::ArgMax && replace cudamalloc … #555
Job | Run time |
---|---|
2h 21m 38s | |
2h 3m 24s | |
2h 1m 25s | |
2h 12m 56s | |
1h 51m 45s | |
2h 42m 34s | |
1h 51m 5s | |
2h 57m 55s | |
0s | |
18h 2m 42s |
Job | Run time |
---|---|
2h 21m 38s | |
2h 3m 24s | |
2h 1m 25s | |
2h 12m 56s | |
1h 51m 45s | |
2h 42m 34s | |
1h 51m 5s | |
2h 57m 55s | |
0s | |
18h 2m 42s |