We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 75473cf commit c4a6228Copy full SHA for c4a6228
python/test/unit/language/test_matmul.py
@@ -339,7 +339,12 @@ def fp8e8m0_to_float32(scale):
339
@pytest.mark.parametrize("nonKDim", ([0, 16, 32] if is_hip_cdna() else [0]))
340
def test_mxfp(M, N, K, BLOCK_M, BLOCK_N, BLOCK_K, NUM_STAGES, nonKDim, NUM_WARPS, device):
341
if is_xpu():
342
- if (BLOCK_M, BLOCK_N, BLOCK_K) in {(128, 128, 64), (128, 64, 128)}:
+ if (nonKDim == 0 and NUM_WARPS == 4 and (M, N, K, BLOCK_M, BLOCK_N, BLOCK_K) in {
343
+ (1024, 512, 256, 128, 64, 128),
344
+ (1024, 512, 256, 128, 128, 64),
345
+ (128, 256, 256, 128, 128, 64),
346
+ (128, 128, 128, 128, 128, 64),
347
+ }):
348
pytest.skip("https://github.com/intel/intel-xpu-backend-for-triton/issues/3677")
349
elif (BLOCK_M, BLOCK_N, BLOCK_K) == (128, 256, 256) and \
350
triton.runtime.driver.active.utils.get_device_properties(
0 commit comments