Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Core] Use os.sched_yield in ShmRingBuffer instead of time.sleep #9994

Merged
merged 1 commit into from
Nov 5, 2024
Merged

[Core] Use os.sched_yield in ShmRingBuffer instead of time.sleep #9994

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 5 additions & 10 deletions vllm/distributed/device_communicators/shm_broadcast.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@
import os
import pickle
import time
from contextlib import contextmanager
Expand All @@ -18,12 +19,6 @@

VLLM_RINGBUFFER_WARNING_INTERVAL = envs.VLLM_RINGBUFFER_WARNING_INTERVAL

# time to wait if the queue is full or empty
# if we sleep for too short, it will consume too much CPU
# if we sleep for too long, it will slow down the writer/reader
# 0.1 us is a good balance
RINGBUFFER_SLEEP_INTERVAL = 1e-7

logger = init_logger(__name__)


Expand Down Expand Up @@ -333,8 +328,8 @@ def acquire_write(self):
# if this block is not ready to write,
# we need to wait until it is read by all readers

# wait for a while
time.sleep(RINGBUFFER_SLEEP_INTERVAL)
# Release the processor to other threads
os.sched_yield()

# if we wait for a long time, we should warn the user
if (time.monotonic() - start_time >
Expand Down Expand Up @@ -387,8 +382,8 @@ def acquire_read(self):
# if this block is not ready,
# we need to wait until it is written

# wait for a while
time.sleep(RINGBUFFER_SLEEP_INTERVAL)
# Release the processor to other threads
os.sched_yield()

# if we wait for a long time, we should warn the user
if (time.monotonic() - start_time >
Expand Down