[XPU][UT] Fix OOM and skip graph case (#49287)

Signed-off-by: mayuyuace <qiming1.zhang@intel.com>
Signed-off-by: Qiming Zhang <qiming1.zhang@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
This commit is contained in:
Qiming Zhang
2026-08-18 09:47:35 +08:00
committed by GitHub
co-authored by Kunshang Ji
parent 69d3335066
commit f4b161d7fc
6 changed files with 46 additions and 21 deletions
@@ -104,9 +104,9 @@ async def async_engine():
from vllm.platforms import current_platform
if current_platform.is_rocm():
from tests.utils import wait_for_rocm_memory_to_settle
from tests.utils import wait_for_memory_to_settle
wait_for_rocm_memory_to_settle(threshold_ratio=1.0 - gpu_memory_utilization)
wait_for_memory_to_settle(threshold_ratio=1.0 - gpu_memory_utilization)
engine_args = AsyncEngineArgs(**ENGINE_CONFIG)
llm = AsyncLLM.from_engine_args(engine_args)
@@ -122,9 +122,9 @@ async def async_engine():
from vllm.distributed import cleanup_dist_env_and_memory
cleanup_dist_env_and_memory()
from tests.utils import wait_for_rocm_memory_to_settle
from tests.utils import wait_for_memory_to_settle
wait_for_rocm_memory_to_settle(threshold_ratio=1.0 - gpu_memory_utilization)
wait_for_memory_to_settle(threshold_ratio=1.0 - gpu_memory_utilization)
def test_voxtral_realtime_forward(audio_assets, tokenizer, vllm_runner, monkeypatch):