Skip to content

Commit db1c91e

Browse files
authored
change k's value to a multiple of 256 (#98)
* Add new problem nvfp4_gemm to nvidia.yaml * change k's value to a multiple of 256 for better perf (simplify some logic). * revert unnecessary change.
1 parent 07f0321 commit db1c91e

1 file changed

Lines changed: 8 additions & 8 deletions

File tree

  • problems/nvidia/nvfp4_group_gemm

problems/nvidia/nvfp4_group_gemm/task.yml

Lines changed: 8 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -45,16 +45,16 @@ templates:
4545
Python: "template.py"
4646

4747
tests:
48-
- {"m": [96, 128], "n": [128, 256], "k": [128, 512], "g": 2, "seed": 1111}
48+
- {"m": [96, 128], "n": [128, 256], "k": [256, 512], "g": 2, "seed": 1111}
4949
- {"m": [256, 72], "n": [512, 384], "k": [256, 256], "g": 2, "seed": 1111}
50-
- {"m": [128, 128], "n": [128, 256], "k": [128, 512], "g": 2, "seed": 1111}
51-
- {"m": [80, 128, 256], "n": [384, 256, 128], "k": [256, 512, 128], "g": 3, "seed": 1111}
52-
- {"m": [64, 72, 96], "n": [128, 384, 512], "k": [384, 512, 128], "g": 3, "seed": 1111}
53-
- {"m": [64, 256, 128], "n": [768, 128, 256], "k": [512, 512, 128], "g": 3, "seed": 1111}
54-
- {"m": [128, 128, 64], "n": [256, 512, 512], "k": [768, 128, 768], "g": 3, "seed": 1111}
55-
- {"m": [128, 128, 128, 128], "n": [128, 128, 128, 128], "k": [128, 128, 128, 128], "g": 4, "seed": 1111}
50+
- {"m": [128, 128], "n": [128, 256], "k": [512, 256], "g": 2, "seed": 1111}
51+
- {"m": [80, 128, 256], "n": [384, 256, 128], "k": [256, 512, 256], "g": 3, "seed": 1111}
52+
- {"m": [64, 72, 96], "n": [128, 384, 512], "k": [512, 512, 256], "g": 3, "seed": 1111}
53+
- {"m": [64, 256, 128], "n": [768, 128, 256], "k": [512, 256, 512], "g": 3, "seed": 1111}
54+
- {"m": [128, 128, 64], "n": [256, 512, 512], "k": [768, 256, 768], "g": 3, "seed": 1111}
55+
- {"m": [128, 128, 128, 128], "n": [128, 128, 128, 128], "k": [512, 256, 512, 256], "g": 4, "seed": 1111}
5656
- {"m": [40, 56, 384, 512], "n": [512, 384, 256, 128], "k": [256, 256, 256, 256], "g": 4, "seed": 1111}
57-
- {"m": [512, 384, 256, 128], "n": [256, 256, 256, 256], "k": [512, 128, 512, 128], "g": 4, "seed": 1111}
57+
- {"m": [512, 384, 256, 128], "n": [256, 256, 256, 256], "k": [512, 768, 512, 768], "g": 4, "seed": 1111}
5858

5959
benchmarks:
6060
- {"m": [80, 176, 128, 72, 64, 248, 96, 160], "n": [4096, 4096, 4096, 4096, 4096, 4096, 4096, 4096], "k": [7168, 7168, 7168, 7168, 7168, 7168, 7168, 7168], "g": 8, "seed": 1111}

0 commit comments

Comments
 (0)