nn.DataParallel and GPU indexes -- Buggy... #2167
Description
Hi, all:
net = torch.nn.DataParallel(model, device_ids=[0]) always fail...
I'm pretty sure I've got ONLY 1 GPU, with the following command line:
jiapei@jiapei-GT72-6QE:~$ sudo lshw -C display
[sudo] password for jiapei:
*-display
description: VGA compatible controller
product: GM204M [GeForce GTX 980M]
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:01:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:128 memory:db000000-dbffffff memory:b0000000-bfffffff memory:c0000000-c1ffffff ioport:e000(size=128) memory:c0000-dffff
jiapei@jiapei-GT72-6QE:~$
And, as you can see, it is already device_ids=[0] already puts GPU in a list...
But, if I try:
net = torch.nn.DataParallel(model, device_ids=[0,1]) , then, succeeded.
However, I ONLY have 1 single GPU ...
Can anybody give me a hand please?
Cheers
Pei