Skip to content

[ROCm] Skip TF32 dot algorithm tests on Navi/RDNA GPUs #783

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

Open
srinivamd wants to merge 1 commit into
base: rocm-jaxlib-v0.9.1
Choose a base branch
from
Open

[ROCm] Skip TF32 dot algorithm tests on Navi/RDNA GPUs #783

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
18 changes: 9 additions & 9 deletions tests/lax_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1180,6 +1180,15 @@ def testDotAlgorithm(self, algorithm, dtype):
raise SkipTest(
f"The dot algorithm '{algorithm}' requires CUDA compute "
"capability >= 8.0.")
# TF32 is not supported on Navi/RDNA GPUs. XLA only allows it on
# MI100+ (gfx9 CDNA) -- see xla/service/algorithm_util.cc.
if (algorithm in {lax.DotAlgorithmPreset.TF32_TF32_F32,
lax.DotAlgorithmPreset.TF32_TF32_F32_X3}
and jtu.is_device_rocm()
and "Radeon" in jax.local_devices()[0].device_kind):
raise SkipTest(
f"The dot algorithm '{algorithm}' is not supported on "
"Navi/RDNA GPUs.")
elif algorithm not in {
lax.DotAlgorithmPreset.DEFAULT,
lax.DotAlgorithmPreset.ANY_F8_ANY_F8_F32,
Expand All @@ -1189,15 +1198,6 @@ def testDotAlgorithm(self, algorithm, dtype):
}:
raise SkipTest(
f"The dot algorithm '{algorithm}' is not supported on GPU.")
if jtu.test_device_matches(["tpu"]):
if algorithm not in {
lax.DotAlgorithmPreset.DEFAULT,
lax.DotAlgorithmPreset.BF16_BF16_F32,
lax.DotAlgorithmPreset.BF16_BF16_F32_X3,
lax.DotAlgorithmPreset.BF16_BF16_F32_X6,
}:
raise SkipTest(
f"The dot algorithm '{algorithm}' is not supported on TPU."
)
lhs_shape = (3, 4)
rhs_shape = (4, 3)
Expand Down