xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 11:45:22 +08:00
Code Issues Packages Projects Releases Wiki Activity

[UX] Rename CUTLASS_MLA_VLLM_V1 to CUTLASS_MLA (#21966)

Browse Source 
Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin 2025-07-30 23:40:34 -04:00 committed by GitHub
parent ec02e536df
commit 61445453df
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/engine/arg_utils.py
View File

@ -1417,7 +1417,7 @@ class EngineArgs:
"PALLAS_VLLM_V1", "PALLAS_VLLM_V1",
"TRITON_ATTN_VLLM_V1", "TRITON_ATTN_VLLM_V1",
"TRITON_MLA", "TRITON_MLA",
"CUTLASS_MLA_VLLM_V1", "CUTLASS_MLA",
"FLASHMLA", "FLASHMLA",
"FLASHINFER", "FLASHINFER",
"FLASHINFER_VLLM_V1", "FLASHINFER_VLLM_V1",

10
vllm/platforms/cuda.py
View File

@ -162,7 +162,7 @@ class CudaPlatformBase(Platform):
if cls.is_device_capability(100): if cls.is_device_capability(100):
# Blackwell => Force CutlassMLA. # Blackwell => Force CutlassMLA.
use_cutlass_mla = True use_cutlass_mla = True
envs.VLLM_ATTENTION_BACKEND = "CUTLASS_MLA_VLLM_V1" envs.VLLM_ATTENTION_BACKEND = "CUTLASS_MLA"
else: else:
# Not Blackwell # Not Blackwell
use_flashmla = True use_flashmla = True
@ -170,7 +170,7 @@ class CudaPlatformBase(Platform):
# Forced case # Forced case
use_flashmla = (envs.VLLM_ATTENTION_BACKEND == "FLASHMLA") use_flashmla = (envs.VLLM_ATTENTION_BACKEND == "FLASHMLA")
use_cutlass_mla = ( use_cutlass_mla = (
envs.VLLM_ATTENTION_BACKEND == "CUTLASS_MLA_VLLM_V1") envs.VLLM_ATTENTION_BACKEND == "CUTLASS_MLA")
from vllm.attention.ops.flashmla import is_flashmla_supported from vllm.attention.ops.flashmla import is_flashmla_supported
if use_flashmla and is_flashmla_supported()[0] \ if use_flashmla and is_flashmla_supported()[0] \
@ -182,7 +182,7 @@ class CudaPlatformBase(Platform):
if use_cutlass_mla and cache_config.block_size != 128: if use_cutlass_mla and cache_config.block_size != 128:
cache_config.block_size = 128 cache_config.block_size = 128
logger.info("Forcing kv cache block size to 128 for " logger.info("Forcing kv cache block size to 128 for "
"CUTLASS_MLA_VLLM_V1 backend.") "CUTLASS_MLA backend.")
compilation_config = vllm_config.compilation_config compilation_config = vllm_config.compilation_config
if (envs.VLLM_ALL2ALL_BACKEND == "deepep_high_throughput" if (envs.VLLM_ALL2ALL_BACKEND == "deepep_high_throughput"
@ -211,9 +211,9 @@ class CudaPlatformBase(Platform):
kv_cache_dtype, block_size, use_v1, kv_cache_dtype, block_size, use_v1,
use_mla) -> str: use_mla) -> str:
if use_mla: if use_mla:
# TODO(lucas): refactor to be more concise # TODO(lucas): refactor to be more concise
# we should probably consider factoring out V1 here # we should probably consider factoring out V1 here
if selected_backend == _Backend.CUTLASS_MLA_VLLM_V1: if selected_backend == _Backend.CUTLASS_MLA:
if use_v1: if use_v1:
logger.info_once("Using Cutlass MLA backend on V1 engine.") logger.info_once("Using Cutlass MLA backend on V1 engine.")
return ("vllm.v1.attention.backends.mla." return ("vllm.v1.attention.backends.mla."

2
vllm/platforms/interface.py
View File

@ -53,7 +53,7 @@ class _Backend(enum.Enum):
TRITON_MLA_VLLM_V1 = enum.auto() TRITON_MLA_VLLM_V1 = enum.auto()
FLASHMLA_VLLM_V1 = enum.auto() FLASHMLA_VLLM_V1 = enum.auto()
FLASHMLA = enum.auto() # Supported by V1 FLASHMLA = enum.auto() # Supported by V1
CUTLASS_MLA_VLLM_V1 = enum.auto() CUTLASS_MLA = enum.auto()
PALLAS = enum.auto() PALLAS = enum.auto()
PALLAS_VLLM_V1 = enum.auto() PALLAS_VLLM_V1 = enum.auto()
IPEX = enum.auto() IPEX = enum.auto()

2
vllm/v1/attention/backends/mla/cutlass_mla.py
View File

@ -21,7 +21,7 @@ class CutlassMLABackend(MLACommonBackend):
@staticmethod @staticmethod
def get_name() -> str: def get_name() -> str:
return "CUTLASS_MLA_VLLM_V1" return "CUTLASS_MLA"
@staticmethod @staticmethod
def get_impl_cls() -> type["CutlassMLAImpl"]: def get_impl_cls() -> type["CutlassMLAImpl"]: