nn.DataParallel(model).cuda() hang

mnist.py

from __future__ import print_function
import argparse
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets, transforms
from torch.autograd import Variable


parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size', type=int, default=64, metavar='N',
                    help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',
                    help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs', type=int, default=10, metavar='N',
                    help='number of epochs to train (default: 10)')
parser.add_argument('--lr', type=float, default=0.01, metavar='LR',
                    help='learning rate (default: 0.01)')
parser.add_argument('--momentum', type=float, default=0.5, metavar='M',
                    help='SGD momentum (default: 0.5)')
parser.add_argument('--no-cuda', action='store_true', default=False,
                    help='disables CUDA training')
parser.add_argument('--seed', type=int, default=1, metavar='S',
                    help='random seed (default: 1)')
parser.add_argument('--log-interval', type=int, default=10, metavar='N',
                    help='how many batches to wait before logging training status')
args = parser.parse_args()
args.cuda = not args.no_cuda 

torch.manual_seed(args.seed)
if args.cuda:
    torch.cuda.manual_seed(args.seed)


kwargs = {'num_workers': 2, 'pin_memory': True} if args.cuda else {}
train_loader = torch.utils.data.DataLoader(
    datasets.MNIST('../data', train=True, download=True,
                   transform=transforms.Compose([
                       transforms.ToTensor(),
                       transforms.Normalize((0.1307,), (0.3081,))
                   ])),
    batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = torch.utils.data.DataLoader(
    datasets.MNIST('../data', train=False, transform=transforms.Compose([
                       transforms.ToTensor(),
                       transforms.Normalize((0.1307,), (0.3081,))
                   ])),
    batch_size=args.batch_size, shuffle=True, **kwargs)


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
        self.conv2_drop = nn.Dropout2d()
        self.fc1 = nn.Linear(320, 50)
        self.fc2 = nn.Linear(50, 10)

    def forward(self, x):
        x = F.relu(F.max_pool2d(self.conv1(x), 2))
        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))


        x = x.view(-1, self.num_flat_features(x))
        x = F.relu(self.fc1(x))
        x = F.dropout(x, training=self.training)
        x = self.fc2(x)
        output = F.log_softmax(x)
        return output

    def num_flat_features(self, x):
        size = x.size()[1:]  # all dimensions except the batch dimension
        num_features = 1
        for s in size:
            num_features *= s
        return num_features

model = Net()
if args.cuda:
    model=torch.nn.DataParallel(model, device_ids=[0,1]).cuda()
   
optimizer = optim.SGD(model.parameters(), lr=args.lr, momentum=args.momentum)

def train(epoch):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        if args.cuda:
            data, target = data.cuda(), target.cuda()
        data, target = torch.autograd.Variable(data), torch.autograd.Variable(target)
        optimizer.zero_grad()
        output = model(data)
        

        loss = F.nll_loss(output, target)

        loss.backward()
        optimizer.step()
        if batch_idx % args.log_interval == 0:
            correct = 0


            pred = output.data.max(1)[1]  # get the index of the max log-probability
            correct += pred.eq(target.data).sum()

            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}\t Accuracy: {}/{} ({:.0f}%)'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                100. * batch_idx / len(train_loader), loss.data[0],
                correct, len(target),
                100. * correct / len(target)))

def test(epoch):
    model.eval()
    test_loss = 0
    correct = 0
    for data, target in test_loader:
        if args.cuda:
            data, target = data.cuda(), target.cuda()
        data, target = Variable(data, volatile=True), Variable(target)
        output = model(data)

        test_loss += F.nll_loss(output, target).data[0]
        pred = output.data.max(1)[1] # get the index of the max log-probability
        correct += pred.eq(target.data).cpu().sum()

    test_loss = test_loss
    test_loss /= len(test_loader) # loss function already averages over batch size
    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))


for epoch in range(1, args.epochs + 1):
    train(epoch)
    test(epoch)

stacktrace

#0  0x00007ffff7bc9827 in futex_abstimed_wait_cancelable (private=0, abstime=0x0, expected=0, futex_word=0x7fff94000c10)
    at ../sysdeps/unix/sysv/linux/futex-internal.h:205
#1  do_futex_wait (sem=sem@entry=0x7fff94000c10, abstime=0x0) at sem_waitcommon.c:111
#2  0x00007ffff7bc98d4 in __new_sem_wait_slow (sem=0x7fff94000c10, abstime=0x0) at sem_waitcommon.c:181
#3  0x00007ffff7bc997a in __new_sem_wait (sem=<optimized out>) at sem_wait.c:29
#4  0x000055555563c516 in PyThread_acquire_lock_timed ()
#5  0x00005555556d63ac in lock_PyThread_acquire_lock ()
#6  0x00005555556616e4 in _PyCFunction_FastCallDict ()
#7  0x00005555556ef4ec in call_function ()
#8  0x0000555555713134 in _PyEval_EvalFrameDefault ()
#9  0x00005555556e8346 in _PyEval_EvalCodeWithName ()
#10 0x00005555556e93b1 in fast_function ()
#11 0x00005555556ef5c5 in call_function ()
#12 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#13 0x00005555556e8346 in _PyEval_EvalCodeWithName ()
#14 0x00005555556e93b1 in fast_function ()
#15 0x00005555556ef5c5 in call_function ()
#16 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#17 0x00005555556e8867 in _PyEval_EvalCodeWithName ()
#18 0x00005555556e93b1 in fast_function ()
#19 0x00005555556ef5c5 in call_function ()
#20 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#21 0x00005555556e917b in fast_function ()
#22 0x00005555556ef5c5 in call_function ()
#23 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#24 0x00005555556e8346 in _PyEval_EvalCodeWithName ()
#25 0x00005555556e9b2d in _PyFunction_FastCallDict ()
#26 0x0000555555661aaf in _PyObject_FastCallDict ()
#27 0x0000555555666563 in _PyObject_Call_Prepend ()
#28 0x00005555556614fe in PyObject_Call ()
#29 0x00005555557148b4 in _PyEval_EvalFrameDefault ()
#30 0x00005555556e8346 in _PyEval_EvalCodeWithName ()
#31 0x00005555556e988f in _PyFunction_FastCallDict ()
#32 0x0000555555661aaf in _PyObject_FastCallDict ()
#33 0x0000555555666563 in _PyObject_Call_Prepend ()
#34 0x00005555556614fe in PyObject_Call ()
#35 0x00005555556c1787 in slot_tp_call ()
#36 0x00005555556618cb in _PyObject_FastCallDict ()
#37 0x00005555556ef63e in call_function ()
#38 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#39 0x00005555556e917b in fast_function ()
#40 0x00005555556ef5c5 in call_function ()
#41 0x0000555555713134 in _PyEval_EvalFrameDefault ()
#42 0x00005555556e9ee8 in PyEval_EvalCodeEx ()
#43 0x00005555556eacac in PyEval_EvalCode ()
#44 0x0000555555767ad4 in run_mod ()
#45 0x0000555555767ed1 in PyRun_FileExFlags ()
#46 0x00005555557680d4 in PyRun_SimpleFileExFlags ()
#47 0x000055555576bb9f in Py_Main ()
#48 0x000055555563371e in main ()