[pytorch learning] torch device

1. torch.device The meaning of representation

torch.device For and on behalf of torch.tensor The object assigned to the device
( To put it simply , Is assigned to you CPU still GPU On , And which one GPU On )

2. torch.device Parameter interpretation of

• torch.device Contains a device type （'cpu’ or ’cuda’ Device type ） And the serial number of the optional device .

• If the equipment serial number does not exist , Is the current device
  for example ,torch.Tensor Build with equipment ’cuda’ The result is the same as ’cuda:X’, among X yes torch.cuda.current_device() Result .

3. structure torch.device Two ways

1. Through strings

>>> torch.device('cpu')
device(type='cpu')

>>> torch.device('cuda:0')
device(type='cuda', index=0)

>>> torch.device('cuda')  # current cuda device
device(type='cuda')

2. In the form of string and device number

>>> torch.device('cpu', 0)
device(type='cpu', index=0)


>>> torch.device('cuda', 0)
device(type='cuda', index=0)