There is no way to use the "GPU" when training

Question

<p dir="auto">Hello, when I was training, I found that the gpu was not in use, but the cpu utilizati

Cloud-Liu · Answer

Hello, I have the same problem. Have you solved it?

JCruan519 · Answer

Hi! I am glad that you could follow our work.
In our "train.py" file, you could se

TWENTY-FOU · Answer

<p dir="auto">Thank you for your reply. I have tried your method above, but it has no effect. I wond

TWENTY-FOU · Answer

I have tried using both PC and server and I am very confused.

JCruan519 · Answer

<p dir="auto">Maybe your pytorch and torchvision are installed as a cpu version. You can check the v

TWENTY-FOU · Answer

我一直都是GPU版本的torch，还是想不通是为什么

TWENTY-FOU · Answer

Have I always been a GPU torch, or can't figure out why

TWENTY-FOU · Answer

没解决不知道怎么解决试了很多办法

TWENTY-FOU · Answer

如果你解决了可以教教我吗？

Cloud-Liu · Answer

我也没解决，我把那行用多GPU的代码注释了，设置了numworkers=16，但是跑的时候显存利用率还是0%，不知道为啥

TWENTY-FOU · Answer

我也没解决，我把那行用多GPU的代码注释了，设置了numworkers=16，但是跑的时候显存利用率还是0%，不知道为啥

JCruan519 · Answer

<p dir="auto">你好！我的代码是在win10环境下写的，因为这篇文章模型的体量较小，所以在Ubuntu或者服务器上并没有进行测试。我在上传代码之前用我本地三台不同的win10进行测试，都是

TWENTY-FOU · Answer

你好！我的代码是在win10环境下写的，因为这篇文章模型的体量较小，所以在Ubuntu或者服务器上并没有进行测试。我在上传代码之前用我本地三台不同

JCruan519 · Answer

这篇论文我实验时使用的是RTX3080，上传代码前还使用了A6000和RTX3090进行测试，都是win10下。

TWENTY-FOU · Answer

那请问您300epoch大概用了多长时间？

JCruan519 · Answer

大约是6~7h吧，上限不会超过半天

icyclv · Answer

感觉主要时间浪费在每轮重复resize了，在跑实验前把图像resize保存下比较好

guyuehhhhh · Answer

我也没解决，我把那行用多GPU的代码注释了，设置了numworkers=16，但是跑的时候显存利用率还是0%，不知道为啥

There is no way to use the "GPU" when training about malunet HOT 18 OPEN