Relative Content

Tag Archive for deep-learningdataparallel

My GPUS doesn’t work properly by using nn.DataPrallel

I’m trying to run my model on multiple GPUs but only one GPU is being used.
I am trying to move it onto my 3 GPUs (1, 2, 3). Another user is using GPU 0, so I am using cuda:1 as the default. Here’s my code:
device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu') model = model.to(device) device_ids = [1, 2, 3] # using GPUs 1, 2, 3 model = nn.DataParallel(model, device_ids=device_ids)
The model should be running on these 3 GPUs, but only cuda:1 is active as shown in the following picture: