RTX3070 support

I used my GTX1060 with cudatoolkit 10 on linux and it was working fine, but after I switched to RTX3070 with the same conda environment as for 1060 it was showing me that GPU is available but my DNN wasn’t leaning. I tried to use new environment with cudatoolkit 11, but now it is showing “cuda is not currently available!” Can you help me with my environment and what version of pytorch currently supports RTX3070?

Here is my new environment package list:

# packages in environment at /home/alex/anaconda3/envs/test:
# Name                    Version                   Build  Channel
_libgcc_mutex             0.1                        main  
blas                      1.0                         mkl  
ca-certificates           2021.1.19            h06a4308_0  
certifi                   2020.12.5        py37h06a4308_0  
cloudpickle               1.6.0                      py_0    anaconda
cudatoolkit               11.0.221             h6bb024c_0  
cycler                    0.10.0                   py37_0    anaconda
cytoolz                   0.11.0           py37h7b6447c_0    anaconda
dask-core                 2.30.0                     py_0    anaconda
decorator                 4.4.2                      py_0    anaconda
freetype                  2.10.4               h5ab3b9f_0  
imageio                   2.9.0                      py_0    anaconda
intel-openmp              2020.2                      254  
jpeg                      9b                   h024ee3a_2  
kiwisolver                1.2.0            py37hfd86e86_0    anaconda
lcms2                     2.11                 h396b838_0  
ld_impl_linux-64          2.33.1               h53a641e_7  
libedit                   3.1.20191231         h14c3975_1  
libffi                    3.3                  he6710b0_2  
libgcc-ng                 9.1.0                hdf63c60_0  
libgfortran-ng            7.3.0                hdf63c60_0    anaconda
libpng                    1.6.37               hbc83047_0  
libstdcxx-ng              9.1.0                hdf63c60_0  
libtiff                   4.1.0                h2733197_1  
libuv                     1.40.0               h7b6447c_0  
lz4-c                     1.9.3                h2531618_0  
matplotlib-base           3.3.1            py37h817c723_0    anaconda
mkl                       2020.2                      256  
mkl-service               2.3.0            py37he8ac12f_0  
mkl_fft                   1.2.0            py37h23d657b_0  
mkl_random                1.1.1            py37h0573a6f_0  
ncurses                   6.2                  he6710b0_1  
networkx                  2.5                        py_0    anaconda
ninja                     1.10.2           py37hff7bd54_0  
numpy                     1.19.2           py37h54aff64_0  
numpy-base                1.19.2           py37hfa32c7d_0  
olefile                   0.46                     py37_0  
openssl                   1.1.1i               h27cfd23_0  
pillow                    8.1.0            py37he98fc37_0  
pip                       20.3.3           py37h06a4308_0  
pyparsing                 2.4.7                      py_0    anaconda
python                    3.7.9                h7579374_0  
python-dateutil           2.8.1                      py_0    anaconda
python_abi                3.7                     1_cp37m    conda-forge
pytorch                   1.7.1           py3.7_cuda11.0.221_cudnn8.0.5_0    pytorch
pywavelets                1.1.1            py37h7b6447c_2    anaconda
pyyaml                    5.3.1            py37h7b6447c_1    anaconda
readline                  8.0                  h7b6447c_0  
scikit-image              0.17.2           py37hdf5156a_0    anaconda
scipy                     1.5.2            py37h0b6359f_0  
setuptools                51.3.3           py37h06a4308_4  
six                       1.15.0           py37h06a4308_0  
sqlite                    3.33.0               h62c20be_0  
tifffile                  2020.10.1        py37hdd07704_2    anaconda
tk                        8.6.10               hbc83047_0  
toolz                     0.11.1                     py_0    anaconda
torchaudio                0.7.2                      py37    pytorch
torchvision               0.8.2                py37_cu110    pytorch
tornado                   6.0.4            py37h7b6447c_1    anaconda
tqdm                      4.56.0             pyhd8ed1ab_0    conda-forge
typing_extensions                    py_0  
wheel                     0.36.2             pyhd3eb1b0_0  
xz                        5.2.5                h7b6447c_0  
yaml                      0.2.5                h7b6447c_0    anaconda
zlib                      1.2.11               h7b6447c_3  
zstd                      1.4.5                h9ceee32_0  

I can’t explain how much pain I suffered setting up RTX 3070 before all the official docs were updated.
You need Cuda 11 to make 30XX family work. And make sure you have the latest drivers, of course.

In short, you have two options:

  1. Create a new conda environment and then literally follow Pytorch install docs conda install pytorch torchvision torchaudio cudatoolkit=11.0 -c pytorch . cudatoolkit will install cuda and cdnn.
  2. Or use the official nvidia docker image NVIDIA NGC
1 Like