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 6, 00:45 UTC
[ROCm][CI] Increase OpenAPI schema test timeouts (#38088)
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
Andreas Karatzas committed
Mar 25, 2026 at 05:06 UTC
04cec4f927cc8d1df85518450d8a48a6ae4f52bf
1 file changed
+5
-2
tests/entrypoints/openai/test_openai_schema.py
+5
-2
@@ -14,14 +14,17 @@ from schemathesis.internal.checks import CheckContext
14
from schemathesis.models import Case
15
from schemathesis.transports.responses import GenericResponse
16
17
+from vllm.platforms import current_platform
18
+
19
from ...utils import RemoteOpenAIServer
20
21
schemathesis.experimental.OPEN_API_3_1.enable()
22
23
MODEL_NAME = "HuggingFaceTB/SmolVLM-256M-Instruct"
24
MAXIMUM_IMAGES = 2
23
-DEFAULT_TIMEOUT_SECONDS: Final[int] = 10
24
-LONG_TIMEOUT_SECONDS: Final[int] = 60
25
+_ROCM_TIMEOUT_MULTIPLIER = 3 if current_platform.is_rocm() else 1
26
+DEFAULT_TIMEOUT_SECONDS: Final[int] = 10 * _ROCM_TIMEOUT_MULTIPLIER
27
+LONG_TIMEOUT_SECONDS: Final[int] = 60 * _ROCM_TIMEOUT_MULTIPLIER
28
29
30
@pytest.fixture(scope="module")