biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Attention] Use has_flashinfer helper (#33177)

Browse Source 
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
Matthew Bonanni
2026-01-27 13:33:17 -05:00
committed by GitHub
parent bd92089d33
commit 1cbccb6dba
1 changed files with 4 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
vllm/model_executor/layers/attention/mla_attention.py
View File

@@ -229,7 +229,7 @@ from vllm.model_executor.layers.quantization.utils.quant_utils import (
get_and_maybe_dequant_weights,
)
from vllm.platforms import current_platform
from vllm.utils.flashinfer import has_nvidia_artifactory
from vllm.utils.flashinfer import has_flashinfer, has_nvidia_artifactory
from vllm.utils.math_utils import cdiv, round_down
from vllm.utils.torch_utils import (
direct_register_custom_op,
@@ -599,13 +599,6 @@ except ImportError:
is_vllm_fa = False
@functools.cache
def flashinfer_available() -> bool:
import importlib.util
return importlib.util.find_spec("flashinfer") is not None
def dynamic_per_batched_tensor_quant(
x: torch.Tensor, dtype: torch.dtype = torch.float8_e4m3fn
):
@@ -824,7 +817,7 @@ def use_flashinfer_prefill() -> bool:
vllm_config = get_current_vllm_config()
if not (
not vllm_config.attention_config.disable_flashinfer_prefill
and flashinfer_available()
and has_flashinfer()
and not vllm_config.attention_config.use_cudnn_prefill
and current_platform.is_device_capability_family(100)
):
@@ -838,7 +831,7 @@ def use_cudnn_prefill() -> bool:
vllm_config = get_current_vllm_config()
return (
flashinfer_available()
has_flashinfer()
and vllm_config.attention_config.use_cudnn_prefill
and current_platform.is_device_capability_family(100)
and has_nvidia_artifactory()
@@ -851,7 +844,7 @@ def use_trtllm_ragged_deepseek_prefill() -> bool:
vllm_config = get_current_vllm_config()
if not (
flashinfer_available
has_flashinfer()
and vllm_config.attention_config.use_trtllm_ragged_deepseek_prefill
and current_platform.is_device_capability_family(100)
):