[Bugfix] Register VLLM_BATCH_INVARIANT in envs.py to fix spurious unknown env var warning (#35007)
Signed-off-by: Ranran <1012869439@qq.com> Signed-off-by: Ranran <hzz5361@psu.edu> Signed-off-by: ran <hzz5361@psu.edu> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
@@ -55,37 +55,37 @@ def _clear_supports_cache():
|
||||
# supports_trtllm_attention
|
||||
|
||||
|
||||
@patch("vllm.utils.flashinfer.vllm_is_batch_invariant", return_value=True)
|
||||
def test_supports_batch_invariant_disables(_mock):
|
||||
@patch("vllm.envs.VLLM_BATCH_INVARIANT", True)
|
||||
def test_supports_batch_invariant_disables():
|
||||
assert supports_trtllm_attention() is False
|
||||
|
||||
|
||||
@patch("vllm.utils.flashinfer.vllm_is_batch_invariant", return_value=False)
|
||||
@patch("vllm.envs.VLLM_BATCH_INVARIANT", False)
|
||||
@patch(
|
||||
"vllm.utils.flashinfer.current_platform.is_device_capability_family",
|
||||
return_value=True,
|
||||
)
|
||||
@patch("vllm.utils.flashinfer.has_nvidia_artifactory", return_value=True)
|
||||
def test_supports_sm100_with_artifactory(_art, _cap, _bi):
|
||||
def test_supports_sm100_with_artifactory(_art, _cap):
|
||||
assert supports_trtllm_attention() is True
|
||||
|
||||
|
||||
@patch("vllm.utils.flashinfer.vllm_is_batch_invariant", return_value=False)
|
||||
@patch("vllm.envs.VLLM_BATCH_INVARIANT", False)
|
||||
@patch(
|
||||
"vllm.utils.flashinfer.current_platform.is_device_capability_family",
|
||||
return_value=False,
|
||||
)
|
||||
def test_supports_non_sm100_platform(_cap, _bi):
|
||||
def test_supports_non_sm100_platform(_cap):
|
||||
assert supports_trtllm_attention() is False
|
||||
|
||||
|
||||
@patch("vllm.utils.flashinfer.vllm_is_batch_invariant", return_value=False)
|
||||
@patch("vllm.envs.VLLM_BATCH_INVARIANT", False)
|
||||
@patch(
|
||||
"vllm.utils.flashinfer.current_platform.is_device_capability_family",
|
||||
return_value=True,
|
||||
)
|
||||
@patch("vllm.utils.flashinfer.has_nvidia_artifactory", return_value=False)
|
||||
def test_supports_sm100_without_artifactory(_art, _cap, _bi):
|
||||
def test_supports_sm100_without_artifactory(_art, _cap):
|
||||
assert supports_trtllm_attention() is False
|
||||
|
||||
|
||||
|
||||
@@ -8,7 +8,7 @@ Run `pytest tests/kernels/moe/test_grouped_topk.py`.
|
||||
import pytest
|
||||
import torch
|
||||
|
||||
import vllm.model_executor.layers.batch_invariant as batch_invariant
|
||||
import vllm.envs as envs
|
||||
from vllm.config import (
|
||||
CompilationConfig,
|
||||
VllmConfig,
|
||||
@@ -69,7 +69,7 @@ def test_grouped_topk(
|
||||
|
||||
with set_current_vllm_config(vllm_config), monkeypatch.context() as m:
|
||||
m.setenv("VLLM_USE_FUSED_MOE_GROUPED_TOPK", "0")
|
||||
m.setattr(batch_invariant, "VLLM_BATCH_INVARIANT", True)
|
||||
m.setattr(envs, "VLLM_BATCH_INVARIANT", True)
|
||||
grouped_topk = GroupedTopk(
|
||||
topk=topk,
|
||||
renormalize=renormalize,
|
||||
|
||||
Reference in New Issue
Block a user