xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 06:35:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Fix ImportError causing by triton (#9493)

Browse Source
This commit is contained in:
Mengqing Cao 2024-11-08 13:08:51 +08:00 committed by GitHub
parent ad39bd640c
commit 7371749d54
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/executor/multiproc_gpu_executor.py
View File

@ -13,12 +13,15 @@ from vllm.executor.multiproc_worker_utils import (ProcessWorkerWrapper,
from vllm.logger import init_logger from vllm.logger import init_logger
from vllm.model_executor.layers.sampler import SamplerOutput from vllm.model_executor.layers.sampler import SamplerOutput
from vllm.sequence import ExecuteModelRequest from vllm.sequence import ExecuteModelRequest
from vllm.triton_utils import maybe_set_triton_cache_manager from vllm.triton_utils.importing import HAS_TRITON
from vllm.utils import (_run_task_with_lock, cuda_device_count_stateless, from vllm.utils import (_run_task_with_lock, cuda_device_count_stateless,
cuda_is_initialized, get_distributed_init_method, cuda_is_initialized, get_distributed_init_method,
get_open_port, get_vllm_instance_id, make_async, get_open_port, get_vllm_instance_id, make_async,
update_environment_variables) update_environment_variables)
if HAS_TRITON:
from vllm.triton_utils import maybe_set_triton_cache_manager
logger = init_logger(__name__) logger = init_logger(__name__)
@ -59,7 +62,7 @@ class MultiprocessingGPUExecutor(DistributedGPUExecutor):
torch.set_num_threads(default_omp_num_threads) torch.set_num_threads(default_omp_num_threads)
# workaround for https://github.com/vllm-project/vllm/issues/6103 # workaround for https://github.com/vllm-project/vllm/issues/6103
if world_size > 1: if HAS_TRITON and world_size > 1:
maybe_set_triton_cache_manager() maybe_set_triton_cache_manager()
# Multiprocessing-based executor does not support multi-node setting. # Multiprocessing-based executor does not support multi-node setting.