[ROCm]Skip test_torchao.py::test_pre_quantized_model on CDNA3 arch (#31905)
Signed-off-by: ZhiweiYan-96 <zhiwei.yan@amd.com>
This commit is contained in:
@@ -6,11 +6,17 @@ import importlib.util
|
|||||||
import pytest
|
import pytest
|
||||||
import torch
|
import torch
|
||||||
|
|
||||||
|
from vllm.platforms import current_platform
|
||||||
|
|
||||||
DTYPE = ["bfloat16"]
|
DTYPE = ["bfloat16"]
|
||||||
|
|
||||||
TORCHAO_AVAILABLE = importlib.util.find_spec("torchao") is not None
|
TORCHAO_AVAILABLE = importlib.util.find_spec("torchao") is not None
|
||||||
|
|
||||||
|
|
||||||
|
@pytest.mark.skipif(
|
||||||
|
current_platform.is_rocm() and current_platform.is_fp8_fnuz(),
|
||||||
|
reason="Only fp8_fnuz supported on CDNA3 architecture",
|
||||||
|
)
|
||||||
@pytest.mark.skipif(not TORCHAO_AVAILABLE, reason="torchao is not available")
|
@pytest.mark.skipif(not TORCHAO_AVAILABLE, reason="torchao is not available")
|
||||||
def test_pre_quantized_model(vllm_runner):
|
def test_pre_quantized_model(vllm_runner):
|
||||||
with vllm_runner(
|
with vllm_runner(
|
||||||
|
|||||||
Reference in New Issue
Block a user