DataParal works for 1 GPU but not for more GPUs

distributed
1

I am working on parallelizing my training using DP, however within a single instance (with 4 GPUs) it it not working when selecting more than 1 GPUs. Following is my code snippet:

def main():
        # Set CUDA_VISIBLE_DEVICES to limit the GPUs used
        if args.num_of_gpu > 0:
                os.environ["CUDA_VISIBLE_DEVICES"] = ",".join(str(i) for i in range(args.num_of_gpu))

        # Check available GPUs
        device = "cuda" if torch.cuda.is_available() else "cpu"
        num_available_gpus = torch.cuda.device_count()
        print(f"Running on {num_available_gpus} GPU(s): {torch.cuda.get_device_name(0) if torch.cuda.is_available() else 'No GPU'}")

        # Check if the GPU is available
        # device = torch.device(args.device) if torch.cuda.is_available() else torch.device("cpu")
        # print(f'Main Selected device: {device}')


        # To store the five fold losses
        cv_valid_losses = []
        cv_valid_mse_losses = []
        cv_test_losses   = []
        for i in range (5):

                # Specify the Model
                if model_name=="x":
                        model = DNN() ## note this DNN is imported
                elif model_name=="enc":
                        model = ED(64)
                elif model_name=="dncnn":
                        model = DnCNN(2,5)
                elif model_name=="dncnn_mod":
                        model = DnCNNmod(2,5)
                elif model_name =="unet": ## this is imported as a model
                        model = UNet(n_channels=2,n_classes=1)

                if num_available_gpus > 1:
                        model = nn.DataParallel(model)
                model.to(device)

here is the error I am getting

RuntimeError: module must have its parameters and buffers on device cuda:0 (device_ids[0]) but found one of them on device: cuda:1```

Please note, I am calling other models in this main.py. In only main.py I am using the DP. I am not sure whats is the problem or how should I solve. Can anyone please help?
2

Could you post the model definition, please?

3

here it is. please have a look

import torch
import torch.nn as nn
from torchinfo import summary
import torch.nn.functional as F
class DoubleConv(nn.Module):
    """(convolution => [BN] => ReLU) * 2"""

    def __init__(self, in_channels, out_channels, mid_channels=None):
        super().__init__()
        if not mid_channels:
            mid_channels = out_channels
        self.double_conv = nn.Sequential(
            nn.Conv2d(in_channels, mid_channels, kernel_size=3, padding=1, bias=False),
            nn.BatchNorm2d(mid_channels),
            nn.ReLU(inplace=True),
            nn.Conv2d(mid_channels, out_channels, kernel_size=3, padding=1, bias=False),
            nn.BatchNorm2d(out_channels),
            nn.ReLU(inplace=True)
        )

    def forward(self, x):
        return self.double_conv(x)

class Down(nn.Module):
    """Downscaling with maxpool then double conv"""

    def __init__(self, in_channels, out_channels):
        super().__init__()
        self.maxpool_conv = nn.Sequential(
            nn.MaxPool2d(2),
            DoubleConv(in_channels, out_channels)
        )

    def forward(self, x):
        return self.maxpool_conv(x)

class Up(nn.Module):
    """Upscaling then double conv"""

    def __init__(self, in_channels, out_channels, bilinear=True, dp = False):
        super().__init__()
        self.dp      = dp
        # if bilinear, use the normal convolutions to reduce the number of channels
        self.dropout = nn.Dropout(0.5)
        if bilinear:
            self.up = nn.Upsample(scale_factor=2, mode='bilinear', align_corners=True)
            self.conv = DoubleConv(in_channels, out_channels, in_channels // 2)
        else:
            self.up = nn.ConvTranspose2d(in_channels, in_channels // 2, kernel_size=2, stride=2)
            self.conv = DoubleConv(in_channels, out_channels)

    def forward(self, x1, x2):
        x1 = self.up(x1)
        if self.dp:
            x1 = self.dropout(x1)
        # input is CHW
        diffY = x2.size()[2] - x1.size()[2]
        diffX = x2.size()[3] - x1.size()[3]

        x1 = F.pad(x1, [diffX // 2, diffX - diffX // 2,
                        diffY // 2, diffY - diffY // 2])
        x = torch.cat([x2, x1], dim=1)
        return self.conv(x)

class OutConv(nn.Module):
    def __init__(self, in_channels, out_channels):
        super(OutConv, self).__init__()
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size=1)

    def forward(self, x):
        return self.conv(x)

class UNet(nn.Module):
    def __init__(self, n_channels, n_classes, bilinear=False):
        super(UNet, self).__init__()
        self.n_channels = n_channels
        self.n_classes = n_classes
        self.bilinear = bilinear

        self.inc = DoubleConv(n_channels, 64)
        self.down1 = Down(64, 128)
        self.down2 = Down(128, 256)
        self.down3 = Down(256, 512)
        factor = 2 if bilinear else 1
        self.down4 = Down(512, 1024 // factor)
        self.up1 = Up(1024, 512 // factor, bilinear)
        self.up2 = Up(512, 256 // factor, bilinear)
        self.up3 = Up(256, 128 // factor, bilinear,dp=True)
        self.up4 = Up(128, 64, bilinear,dp=True)
        self.outc = OutConv(64, n_classes)
        self.dropout = nn.Dropout2d(0.5)

    def forward(self, x):
        x1 = self.inc(x)
        x1 = self.dropout(x1)
        x2 = self.down1(x1)
        x2 = self.dropout(x2)
        x3 = self.down2(x2)
        x3 = self.dropout(x3)
        # x4 = self.down3(x3)
        # x5 = self.down4(x4)
        # x = self.up1(x5, x4)
        # x = self.up2(x, x3)
        x = self.up3(x3, x2)

        x = self.up4(x, x1)
        logits = self.outc(x)
        return logits

if __name__=="__main__":
    model = UNet(n_channels=3,n_classes=1)
    summary(model, input_size=(128,3, 10, 10))
    print("Done!")
4

Thanks for the code! Your code works fine for me:

if __name__=="__main__":
    model = UNet(n_channels=3,n_classes=1)
    model = nn.DataParallel(model).cuda()
    x = torch.randn(128,3, 10, 10).cuda()
    out = model(x)
    print(out.shape)

and the proper output shape is shown without any errors being raised.

1 Like