EngineX-Ascend/enginex-ascend-910-vllm
10
0
Fork 1
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init v0.11.0rc0

Browse Source
This commit is contained in:
luopingyi
2025-10-14 10:38:28 +08:00
parent 67afd0ea78
commit 66dc16f966
278 changed files with 28130 additions and 11708 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
vllm_ascend/envs.py
View File

@@ -131,6 +131,26 @@ env_variables: Dict[str, Callable[[], Any]] = {
# this feature is supported in A2, and eager mode will get better performance.
"VLLM_ASCEND_ENABLE_MATMUL_ALLREDUCE":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_MATMUL_ALLREDUCE", '0'))),
# Whether to enable FlashComm optimization when tensor parallel is enabled.
# This feature will get better performance when concurrency is large.
"VLLM_ASCEND_ENABLE_FLASHCOMM":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_FLASHCOMM", '0'))),
# Whether to enable MLP weight prefetch, only used in small concurrency.
"VLLM_ASCEND_ENABLE_PREFETCH_MLP":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_PREFETCH_MLP", '0'))),
# buffer size for gate up prefetch
"VLLM_ASCEND_MLP_GATE_UP_PREFETCH_SIZE":
lambda: int(
os.getenv("VLLM_ASCEND_MLP_GATE_UP_PREFETCH_SIZE", 18 * 1024 * 1024)),
# buffer size for down proj prefetch
"VLLM_ASCEND_MLP_DOWN_PREFETCH_SIZE":
lambda: int(
os.getenv("VLLM_ASCEND_MLP_DOWN_PREFETCH_SIZE", 18 * 1024 * 1024)),
# Whether to enable dense model and general optimizations for better performance.
# Since we modified the base parent class `linear`, this optimization is also applicable to other model types.
# However, there might be hidden issues, and it is currently recommended to prioritize its use with dense models.
"VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE", '0'))),
# Whether to enable mlp optimize when tensor parallel is enabled.
# this feature in eager mode will get better performance.
"VLLM_ASCEND_ENABLE_MLP_OPTIMIZE":
@@ -139,11 +159,16 @@ env_variables: Dict[str, Callable[[], Any]] = {
# caused by the initialization of the Mooncake connector.
"PHYSICAL_DEVICES":
lambda: os.getenv("PHYSICAL_DEVICES", None),
# Whether to enable msMonitor tool to monitor the performance of vllm-ascend.
"MSMONITOR_USE_DAEMON":
lambda: bool(int(os.getenv("MSMONITOR_USE_DAEMON", '0'))),
# Timeout (in seconds) for delayed KVCache block release. In the prefill
# node, if a request is marked for delayed KV block release and the blocks
# are not freed within this timeout, they will be forcibly released.
"VLLM_ASCEND_KVCACHE_DELAY_FREE_TIMEOUT":
lambda: int(os.getenv("VLLM_ASCEND_KVCACHE_DELAY_FREE_TIMEOUT", 250)),
"VLLM_ASCEND_ENABLE_MLAPO":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_MLAPO", '0'))),
}
# end-env-vars-definition
@@ -157,4 +182,4 @@ def __getattr__(name: str):
def __dir__():
return list(env_variables.keys())
return list(env_variables.keys())