HongChuang/ultralytics-ascend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 3

ultralytics 8.2.12 revert os.sched_setaffinity() to fix train slowdowns (#11859)

Browse source 
Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
This commit is contained in:
Laughing 2024-05-11 00:20:27 +08:00 • committed by GitHub
parent 94f77368a0
commit 65ecf1442e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 2 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ultralytics/__init__.py
View file

@ -1,6 +1,6 @@
# Ultralytics YOLO 🚀, AGPL-3.0 license # Ultralytics YOLO 🚀, AGPL-3.0 license
__version__ = "8.2.11" __version__ = "8.2.12"
from ultralytics.data.explorer.explorer import Explorer from ultralytics.data.explorer.explorer import Explorer
from ultralytics.models import RTDETR, SAM, YOLO, YOLOWorld from ultralytics.models import RTDETR, SAM, YOLO, YOLOWorld

4
ultralytics/data/build.py
View file

@ -21,7 +21,7 @@ from ultralytics.data.loaders import (
autocast_list, autocast_list,
) )
from ultralytics.data.utils import IMG_FORMATS, PIN_MEMORY, VID_FORMATS from ultralytics.data.utils import IMG_FORMATS, PIN_MEMORY, VID_FORMATS
from ultralytics.utils import LINUX, NUM_THREADS, RANK, colorstr from ultralytics.utils import LINUX, RANK, colorstr
from ultralytics.utils.checks import check_file from ultralytics.utils.checks import check_file
@ -79,8 +79,6 @@ def seed_worker(worker_id): # noqa
worker_seed = torch.initial_seed() % 2**32 worker_seed = torch.initial_seed() % 2**32
np.random.seed(worker_seed) np.random.seed(worker_seed)
random.seed(worker_seed) random.seed(worker_seed)
if LINUX and hasattr(os, "sched_setaffinity"): # unsupported on macOS and Windows
os.sched_setaffinity(0, range(NUM_THREADS)) # fix https://github.com/ultralytics/ultralytics/pull/11195
def build_yolo_dataset(cfg, img_path, batch, data, mode="train", rect=False, stride=32, multi_modal=False): def build_yolo_dataset(cfg, img_path, batch, data, mode="train", rect=False, stride=32, multi_modal=False):