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 15, 06:00 UTC
[Misc] Cleanup compilation tests (#19343)
Signed-off-by: rzou <zou3519@gmail.com>
Richard Zou committed
Jun 9, 2025 at 03:05 UTC
3a4d41770707f90d1d65e5c4a4c5cfb0f633dd2a
1 file changed
-9
tests/compile/test_config.py
-9
@@ -11,15 +11,6 @@ from vllm.config import (CompilationConfig, CompilationLevel, VllmConfig,
11
from .piecewise.test_simple import SillyModel
12
13
14
-@pytest.fixture(scope="function", autouse=True)
15
-def use_v1(monkeypatch):
16
- """
17
- TODO(rzou): The rest of tests/compile runs VLLM_USE_V1=0 right now,
18
- I'll switch them over later.
19
- """
20
- monkeypatch.setenv('VLLM_USE_V1', '1')
21
-
22
-
14
@pytest.mark.parametrize("enabled", [True, False])
15
def test_use_cudagraphs(enabled):
16
assert vllm.envs.VLLM_USE_V1