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

[core][executor] simplify instance id (#10976)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
youkaichao
2024-12-07 09:33:45 -08:00
committed by GitHub
parent 78029b34ed
commit 1b62745b1d
10 changed files with 22 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/executor/multiproc_gpu_executor.py
View File

@@ -16,7 +16,7 @@ from vllm.sequence import ExecuteModelRequest
from vllm.triton_utils.importing import HAS_TRITON
from vllm.utils import (_run_task_with_lock, cuda_device_count_stateless,
cuda_is_initialized, get_distributed_init_method,
get_open_port, get_vllm_instance_id, make_async,
get_open_port, make_async,
update_environment_variables)
if HAS_TRITON:
@@ -37,9 +37,6 @@ class MultiprocessingGPUExecutor(DistributedGPUExecutor):
world_size = self.parallel_config.world_size
tensor_parallel_size = self.parallel_config.tensor_parallel_size
# Ensure that VLLM_INSTANCE_ID is set, to be inherited by workers
os.environ["VLLM_INSTANCE_ID"] = get_vllm_instance_id()
# Disable torch async compiling which won't work with daemonic processes
os.environ["TORCHINDUCTOR_COMPILE_THREADS"] = "1"