Skip to content

Fix split-K test compatibility on M5, M6 #4677

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zcbenz merged 1 commit into from
Oct 12, 2026
+3 −1
Merged

Fix split-K test compatibility on M5, M6 #4677

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion python/tests/test_quantized.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -665,8 +665,10 @@ def test_qmm_splitk_precision(self):
if mx.default_device() == mx.cpu:
self.skipTest("Covers GPU kernels only")
K, N = 2560, 1024
# Keep M at or above get_qmv_batch_limit on all devices. Below it the
# qmv kernel runs, which dequantizes in float32 and not in the input type.
for mode, dtype, M in product(
["affine", "mxfp4"], [mx.bfloat16, mx.float16], [16, 33, 65]
["affine", "mxfp4"], [mx.bfloat16, mx.float16], [25, 33, 65]
):
with self.subTest(mode=mode, dtype=dtype, M=M):
k1, k2 = mx.random.split(mx.random.key(0))
Expand Down
Loading