Fix `select_device()` for Multi-GPU (#6434)

utils/datasets.py

@@ -29,13 +29,13 @@ from tqdm import tqdm
 from utils.augmentations import Albumentations, augment_hsv, copy_paste, letterbox, mixup, random_perspective
 from utils.general import (LOGGER, NUM_THREADS, check_dataset, check_requirements, check_yaml, clean_str,
                            segments2boxes, xyn2xy, xywh2xyxy, xywhn2xyxy, xyxy2xywhn)
-from utils.torch_utils import torch_distributed_zero_first
+from utils.torch_utils import device_count, torch_distributed_zero_first
 
 # Parameters
 HELP_URL = 'https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data'
 IMG_FORMATS = ['bmp', 'dng', 'jpeg', 'jpg', 'mpo', 'png', 'tif', 'tiff', 'webp']  # include image suffixes
 VID_FORMATS = ['asf', 'avi', 'gif', 'm4v', 'mkv', 'mov', 'mp4', 'mpeg', 'mpg', 'wmv']  # include video suffixes
-DEVICE_COUNT = max(torch.cuda.device_count(), 1)
+DEVICE_COUNT = max(device_count(), 1)  # number of CUDA devices
 
 # Get orientation exif tag
 for orientation in ExifTags.TAGS.keys():

utils/torch_utils.py

@@ -53,6 +53,15 @@ def git_describe(path=Path(__file__).parent):  # path must be a directory
         return ''  # not a git repository
 
 
+def device_count():
+    # Returns number of CUDA devices available. Safe version of torch.cuda.device_count().
+    try:
+        cmd = 'nvidia-smi -L | wc -l'
+        return int(subprocess.run(cmd, shell=True, capture_output=True, check=True).stdout.decode().split()[-1])
+    except Exception as e:
+        return 0
+
+
 def select_device(device='', batch_size=0, newline=True):
     # device = 'cpu' or '0' or '0,1,2,3'
     s = f'YOLOv5 🚀 {git_describe() or date_modified()} torch {torch.__version__} '  # string
@@ -61,10 +70,10 @@ def select_device(device='', batch_size=0, newline=True):
     if cpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = '-1'  # force torch.cuda.is_available() = False
     elif device:  # non-cpu device requested
-        nd = torch.cuda.device_count()  # number of CUDA devices
-        assert torch.cuda.is_available(), 'CUDA is not available, use `--device cpu` or do not pass a --device'
+        nd = device_count()  # number of CUDA devices
         assert nd > int(max(device.split(','))), f'Invalid `--device {device}` request, valid devices are 0 - {nd - 1}'
-        os.environ['CUDA_VISIBLE_DEVICES'] = device  # set environment variable (must be after asserts)
+        os.environ['CUDA_VISIBLE_DEVICES'] = device  # set environment variable - must be before assert is_available()
+        assert torch.cuda.is_available(), 'CUDA is not available, use `--device cpu` or do not pass a --device'
 
     cuda = not cpu and torch.cuda.is_available()
     if cuda: