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 10, 12:45 UTC
[CI/Build] Enable test_kv_cache_events_dp for AMD (#31834)
Signed-off-by: Ryan Rock <ryan.rock@amd.com>
Ryan Rock committed
Jan 8, 2026 at 03:00 UTC
8cbdc7eb941e8f3085920b57211ca45df5d71c10
1 file changed
+5
-2
tests/v1/engine/test_engine_core_client.py
+5
-2
@@ -37,8 +37,11 @@ from vllm.v1.executor.abstract import Executor
37
from ...distributed.conftest import MockSubscriber
38
from ...utils import create_new_process_for_each_test
39
40
-if not current_platform.is_cuda():
41
- pytest.skip(reason="V1 currently only supported on CUDA.", allow_module_level=True)
40
+if not current_platform.is_cuda_alike():
41
+ pytest.skip(
42
+ reason="V1 currently only supported on CUDA-alike platforms.",
43
+ allow_module_level=True,
44
+ )
45
46
MODEL_NAME = "meta-llama/Llama-3.2-1B-Instruct"
47
TOKENIZER = AutoTokenizer.from_pretrained(MODEL_NAME)