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

[CI][AMD][BugFix] Skip tests in test_unquantized_backend_selection that should not run on ROCm (#34655)

Browse Source 
Signed-off-by: Randall Smith <Randall.Smith@amd.com>
This commit is contained in:
rasmith
2026-02-18 14:00:40 -06:00
committed by GitHub
parent 95be2a7f22
commit fcd6ac97ed
1 changed files with 7 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
tests/kernels/moe/test_unquantized_backend_selection.py
View File

@@ -9,6 +9,7 @@ from vllm.model_executor.layers.fused_moe.oracle.unquantized import (
UnquantizedMoeBackend,
select_unquantized_moe_backend,
)
from vllm.platforms import current_platform
@pytest.mark.parametrize(
@@ -65,6 +66,9 @@ def test_select_default_backend_by_platform(
"vllm.model_executor.layers.fused_moe.oracle.unquantized.is_supported_config_trtllm_bf16",
return_value=(True, None),
)
@pytest.mark.skipif(
not current_platform.is_cuda(), reason="Only supported on NVIDIA platforms."
)
def test_select_cuda_flashinfer_trtllm_backend(
mock_has_flashinfer, mock_is_supported_trtllm, monkeypatch
):
@@ -101,6 +105,9 @@ def test_select_cuda_flashinfer_trtllm_backend(
"vllm.model_executor.layers.fused_moe.oracle.unquantized.is_supported_config_trtllm_bf16",
return_value=(False, None),
)
@pytest.mark.skipif(
not current_platform.is_cuda(), reason="Only supported on NVIDIA platforms."
)
def test_select_cuda_flashinfer_cutlass_backend(
mock_has_flashinfer, mock_is_supported_trtllm, monkeypatch
):