one machine, two gpus.

[WindSmileValley]

some strange questions, 2 cases to illustrate:
case1:
sei_trainer = Trainer(
gpus=2,
strategy='dp',)
self.model = torch.nn.DataParallel(self.model,device_ids = [0,1]) # it is in a LightningModule class

report mistake: "RuntimeError: Expected tensor for argument #1 'input' to have the same device as tensor for argument #2 'weight'; but device 1 does not equal 0 (while checking arguments for cudnn_convolution)"

however,
if
case1:
sei_trainer = Trainer(
gpus=1,
strategy='dp',)
self.model = torch.nn.DataParallel(self.model,device_ids = [0,1])

then, the code run in 2 gpus.

So, what is happen and how I to fix it?
Thank you all.