A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 12, 16:00 UTC
[CI/Build][AMD] Skip test_multi_shared_storage_connector_consistency in test_multi_connector.py due to hipErrorLaunchFailure when calling .cpu() (#29253)
Signed-off-by: Randall Smith <ransmith@amd.com> Co-authored-by: Randall Smith <ransmith@amd.com>
rasmith committed
Nov 22, 2025 at 22:42 UTC
71362ffab4ef3f5f99ecae2f345f8c689e903f7d
1 file changed
+8
tests/v1/kv_connector/unit/test_multi_connector.py
+8
@@ -20,6 +20,7 @@ from vllm.distributed.kv_transfer.kv_connector.v1.multi_connector import (
20
from vllm.distributed.kv_transfer.kv_connector.v1.nixl_connector import (
21
NixlKVConnectorStats,
22
)
23
+from vllm.platforms import current_platform
24
25
MODEL_NAME = "meta-llama/Llama-3.2-1B-Instruct"
26
@@ -69,6 +70,13 @@ def _compare_directories(dir1: Path, dir2: Path) -> bool:
70
return True
71
72
73
+@pytest.mark.skipif(
74
+ current_platform.is_rocm(),
75
+ reason=(
76
+ "hipErrorLaunchFailure when running this test, see issue:"
77
+ "https://github.com/ROCm/pytorch/issues/2822"
78
+ ),
79
+)
80
def test_multi_shared_storage_connector_consistency():
81
"""
82
Tests that MultiConnector with two SharedStorageConnectors saves