biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Kernel] Tuned FP8 Kernels for Ada Lovelace (#6677)

Browse Source 
Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>
This commit is contained in:
Varun Sundar Rabindranath
2024-07-29 11:42:35 -04:00
committed by GitHub
parent 7cbd9ec7a9
commit 766435e660
6 changed files with 877 additions and 490 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/kernels/test_cutlass.py
View File

@@ -106,8 +106,8 @@ def cutlass_int8_gemm_helper(m: int,
assert torch.allclose(out, baseline, rtol=1e-1, atol=1e0)
@pytest.mark.parametrize("m", [512, 222, 100, 33, 1])
@pytest.mark.parametrize("n", [2048, 256, 1024])
@pytest.mark.parametrize("m", [1, 16, 32, 64, 128, 256, 512, 222, 100, 33])
@pytest.mark.parametrize("n", [2048, 4096, 8192, 16384, 24576, 256, 1024])
@pytest.mark.parametrize("k", [128, 496, 1024])
@pytest.mark.parametrize("per_act_token", [True, False])
@pytest.mark.parametrize("per_out_ch", [True, False])