Pytorch Issue -- Ubuntu 22.04.1 LTS, Release: 22.04

[jjherlan]

Hello and Happy New Year!

I am running:
Ubuntu 22.04.1 LTS, Release: 22.04

I have successfully installed CUDA v. 11.7.0-1 or 11.5 (below)?

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Thu_Nov_18_09:45:30_PST_2021
Cuda compilation tools, release 11.5, V11.5.119
Build cuda_11.5.r11.5/compiler.30672275_0

But I get the following error message during installation:

Found NVCC version: 11.5
nvcc version installed not supported by pytorch!!
Something is wrong with NVCC. Do you want to install the CPU version of pythorch? [Y/n]

Is there a solution to install the GPU version of pytorch yet?

Thank you, in advance.

James

[cappelletto]

Hi James, did you manage to solve this?
I had a similar issue when upgrading to Ubuntu 22.04. Could you post your current environment setup?

pip list

Cheers,
J

[jjherlan]

Sorry to be MIA, just getting back to this project:

pip list

absl-py 1.4.0
affine 2.4.0
albumentations 1.3.1
astunparse 1.6.3
attrs 23.1.0
boltons 23.0.0
brotlipy 0.7.0
cachetools 5.3.1
certifi 2023.5.7
cffi 1.15.1
charset-normalizer 2.0.4
click 8.1.3
click-plugins 1.1.1
cligj 0.7.2
conda 23.5.0
conda-content-trust 0.1.3
conda-package-handling 2.0.2
conda_package_streaming 0.7.0
contourpy 1.1.0
cryptography 39.0.1
cycler 0.11.0
filelock 3.9.0
flatbuffers 23.5.26
fonttools 4.40.0
fsspec 2023.4.0
gast 0.4.0
GDAL 3.4.1
google-auth 2.21.0
google-auth-oauthlib 1.0.0
google-pasta 0.2.0
grpcio 1.56.0
h5py 3.9.0
idna 3.4
imageio 2.31.1
jax 0.4.13
Jinja2 3.1.2
joblib 1.3.1
jsonpatch 1.32
jsonpointer 2.1
keras 2.12.0
kiwisolver 1.4.4
lazy_loader 0.3
libclang 16.0.0
Markdown 3.4.3
MarkupSafe 2.1.2
matplotlib 3.7.2
ml-dtypes 0.2.0
mpmath 1.2.1
networkx 3.0rc1
numpy 1.22.3
oauthlib 3.2.2
opencv-python 4.8.0.74
opencv-python-headless 4.8.0.74
opt-einsum 3.3.0
packaging 23.0
pandas 2.0.3
Pillow 9.3.0
pip 23.1.2
pluggy 1.0.0
protobuf 4.23.3
pyasn1 0.5.0
pyasn1-modules 0.3.0
pycocotools 2.0.6
pycosat 0.6.4
pycparser 2.21
pyOpenSSL 23.0.0
pyparsing 3.0.9
PyQt5 5.15.9
PyQt5-Qt5 5.15.2
PyQt5-sip 12.12.1
PySocks 1.7.1
python-dateutil 2.8.2
pytorch-triton 2.1.0+440fd1bf20
pytz 2023.3
PyWavelets 1.4.1
PyYAML 6.0
qudida 0.0.4
rasterio 1.3.8
requests 2.28.1
requests-oauthlib 1.3.1
rsa 4.9
ruamel.yaml 0.17.21
ruamel.yaml.clib 0.2.6
scikit-image 0.21.0
scikit-learn 1.3.0
scipy 1.11.1
setuptools 68.0.0
shapely 2.0.1
six 1.16.0
sklearn 0.0.post5
snuggs 1.4.7
sympy 1.11.1
tensorboard 2.12.3
tensorboard-data-server 0.7.1
tensorflow 2.12.0
tensorflow-estimator 2.12.0
tensorflow-io-gcs-filesystem 0.32.0
termcolor 2.3.0
threadpoolctl 3.1.0
tifffile 2023.7.4
toolz 0.12.0
torch 2.1.0.dev20230701+cu121
torchaudio 2.1.0.dev20230702+cu121
torchvision 0.16.0.dev20230702+cu121
tqdm 4.65.0
typing_extensions 4.4.0
tzdata 2023.3
urllib3 1.26.15
Werkzeug 2.3.6
wheel 0.40.0
wrapt 1.14.1
zstandard 0.19.0

[alemuntoni]

Hello everyone,
could you please try to type nvcc --version in a terminal and paste the output here?
It seems that the default installation of cuda is 11.5, which is not supported. Maybe the 11.7 installation is just hidden by 11.5.

[jjherlan]

Thank you for the reply. I now have Cuda 11.8. I do not think that is supported yet, could you confirm?

nvcc --version

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0