[CI/Build][AMD] Skip quantization kernels tests that require CUTLASS or e4m3fn when not supported by platform (#30020)

Signed-off-by: Randall Smith <ransmith@amd.com>
Co-authored-by: Randall Smith <ransmith@amd.com>
This commit is contained in:
rasmith
2025-12-09 20:28:37 -06:00
committed by GitHub
parent c3487aca34
commit b75f826fca
3 changed files with 20 additions and 3 deletions

View File

@@ -21,6 +21,9 @@ from vllm.model_executor.layers.quantization.utils.quant_utils import (
from vllm.platforms import current_platform
from vllm.scalar_type import ScalarType, scalar_types
if not current_platform.is_cuda():
pytest.skip("These tests use CUTLASS which requires CUDA", allow_module_level=True)
# TODO: in future PR refactor this and `is_quant_method_supported` in the kernel
# unit tests to a common utility function. Currently the use of
# `is_quant_method_supported` conflates kernels with quantization methods