xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-13 19:55:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[misc] skip p2p check by default (#21904)

Browse Source
This commit is contained in:
youkaichao 2025-07-30 22:05:04 +08:00 committed by GitHub
parent bf668b5bf5
commit e91d3c9cda
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 7 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/envs.py
View File

@ -668,12 +668,14 @@ environment_variables: dict[str, Callable[[], Any]] = {
(os.environ.get("VLLM_ALLOW_RUNTIME_LORA_UPDATING", "0").strip().lower() in (os.environ.get("VLLM_ALLOW_RUNTIME_LORA_UPDATING", "0").strip().lower() in
("1", "true")), ("1", "true")),
# By default, vLLM will check the peer-to-peer capability itself, # We assume drivers can report p2p status correctly.
# in case of broken drivers. See https://github.com/vllm-project/vllm/blob/a9b15c606fea67a072416ea0ea115261a2756058/vllm/distributed/device_communicators/custom_all_reduce_utils.py#L101-L108 for details. # noqa # If the program hangs when using custom allreduce,
# If this env var is set to 1, vLLM will skip the peer-to-peer check, # potantially caused by a bug in the driver (535 series),
# and trust the driver's peer-to-peer capability report. # if might be helpful to set VLLM_SKIP_P2P_CHECK=0
# so that vLLM can verify if p2p is actually working.
# See https://github.com/vllm-project/vllm/blob/a9b15c606fea67a072416ea0ea115261a2756058/vllm/distributed/device_communicators/custom_all_reduce_utils.py#L101-L108 for details. # noqa
"VLLM_SKIP_P2P_CHECK": "VLLM_SKIP_P2P_CHECK":
lambda: os.getenv("VLLM_SKIP_P2P_CHECK", "0") == "1", lambda: os.getenv("VLLM_SKIP_P2P_CHECK", "1") == "1",
# List of quantization kernels that should be disabled, used for testing # List of quantization kernels that should be disabled, used for testing
# and performance comparisons. Currently only affects MPLinearKernel # and performance comparisons. Currently only affects MPLinearKernel