Need help with CUDA_VERSION

vision
1

UserWarning:
Found GPU0 Tesla V100-SXM2-16GB which requires CUDA_VERSION >= 9000 for
optimal performance and fast startup time, but your PyTorch was compiled
with CUDA_VERSION 8000. Please install the correct PyTorch binary
using instructions from http://pytorch.org
…
…
RuntimeError: CUDNN_STATUS_MAPPING_ERROR

But i did install pytorch for CUDA 9.0 and Python 3.5…

Debian 4.9.88-1+deb9u1
Cuda compilation tools, release 9.1, V9.1.85
cuDNN 7.1

2

How did you install PyTorch?
Did you compile from source or install the pre-built binaries?
You can find the install command on the website.

This should be the command you are looking for:

conda install pytorch torchvision cuda90 -c pytorch
2 Likes
3

I using this commands

image
image1124×535 30.6 KB

I double checked for torch uninstalled from python2.7 and python3.5 and after that apply next commands:
pip3 install http://download.pytorch.org/whl/cu90/torch-0.4.0-cp35-cp35m-linux_x86_64.whl
pip3 install torchvision

but this not helps

Thanks for advice, i’ll try using anaconda now

4

After installation, could you check the current CUDA version used in PyTorch with torch.version.cuda?

2 Likes
5

Before that i checked version
image

this is strange

now i’ll try to using anaconda…

6

Ok, now it’s looking good
image

Thank you!

But now i try to launch deepvoice3_pytorch/train.py
and got this error

`ModuleNotFoundError: No module named 'inflect'`

I did try install this module from anaconda but no luck

Solving environment: failed
PackagesNotFoundError: The following packages are not available from current channels:
  - inflect
Current channels:
  - https://repo.anaconda.com/pkgs/main/linux-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/free/linux-64
  - https://repo.anaconda.com/pkgs/free/noarch
  - https://repo.anaconda.com/pkgs/r/linux-64
  - https://repo.anaconda.com/pkgs/r/noarch
  - https://repo.anaconda.com/pkgs/pro/linux-64
  - https://repo.anaconda.com/pkgs/pro/noarch
7

If it’s not in the conda channels, you could try to install it via pip.
Usually conda finds the packages (although you should install it via conda if possible).

8

@ToxaDev How do you install cuda90 pytorch? could you share your solution, thanks.

9

Hi I’m using google colab with my project whichi needs torch<=1.2, but each time i installed the corrct version the colab will need to restartsession and the torch.version.cuda will again be 9.0. how can i fix it?