some errors with windows10 with RTX-3070
Closed this issue · 2 comments
Describe the bug
I'm new to this and I met these errors. I already installed required libraries and cleared the cache.
I run the train.py with a customer dataset zip file and I keep get 'No module named 'upfirdn2d_plugin' warning.
But it seems still running since the memory of GPU is highly occupied.
Then comes 'Evaluating metrics...' I then got this error:
'RuntimeError: [enforce fail at ..\caffe2\serialize\inline_container.cc:145] . PytorchStreamReader failed reading zip archive: failed finding central directory'
Screenshots
Desktop (please complete the following information):
- OS: [Windows 10]
- PyTorch version (pytorch 1.7.1)
- CUDA toolkit version (CUDA 11.0) [specifically it shows like this in pip list: 'torch 1.7.1+cu110']
- NVIDIA driver version 8.1.940.0
- GPU [ RTX 3070]
- Docker: did you use Docker? no
Here is my log
log.txt
probably needs cuda 11.1 since the 3090 requires it, also you will probably not have enough vram to run as recommended.
edit: yeh, log line 11 shows cuda library isnt detected or installed properly, try installing cuda 11.1 from nvidia's dev portal.