Cudnn_status_not_initialized

The error is shown below:

File “/home/zhengyun.zy/anaconda/lib/python2.7/site-packages/torch/nn/modules/module.py”, line 206, in call
result = self.forward(*input, **kwargs)
File “/home/zhengyun.zy/anaconda/lib/python2.7/site-packages/torch/nn/modules/conv.py”, line 237, in forward
self.padding, self.dilation, self.groups)
File “/home/zhengyun.zy/anaconda/lib/python2.7/site-packages/torch/nn/functional.py”, line 40, in conv2d
return f(input, weight, bias)
RuntimeError: CUDNN_STATUS_NOT_INITIALIZED

cudnn is correctly installed

import torch
print(torch.backends.cudnn.is_acceptable(torch.cuda.FloatTensor(1)))
True
print(torch.backends.cudnn.version())
6021

According to the manual, this error is returned when no compatible GPU is found, CUDA driver is not installed or disabled, CUDA runtime API initialization failed.

Any update on this? I’m experiencing the same problem with CUDA 9.1 and cuDNN 7

Had the same issue, resolved after I replaced cuda 9.1 with cuda 9.0.
Cudnn 7, pytorch 0.3.