1

I followed this link here to download tensorflow with gpu support on my Ubuntu 20.04 OS. Here is a snapshot of my .bashrc that I suspect may be the problem but I cannot verify this.

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/home/weiss/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
    eval "$__conda_setup"
else
    if [ -f "/home/weiss/anaconda3/etc/profile.d/conda.sh" ]; then
        . "/home/weiss/anaconda3/etc/profile.d/conda.sh"
    else
        export PATH="/home/weiss/anaconda3/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda initialize <<<

export LD_LIBRARY_PATH=/usr/lib/cuda/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/lib/cuda/include:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/local/cuda-11.0/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
export PATH=/usr/local/cuda-11.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-11.0/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/local/cuda-11.0/include:$LD_LIBRARY_PATH
echo 'export LD_LIBRARY_PATH=/usr/lib/cuda/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc
    echo 'export LD_LIBRARY_PATH=/usr/lib/cuda/include:$LD_LIBRARY_PATH' >> ~/.bashrc


export LD_LIBRARY_PATH=/usr/lib/cuda/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/lib/cuda/include:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/lib/cuda/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/lib/cuda/include:$LD_LIBRARY_PATH

When I run this command

 tf.config.list_physical_devices('GPU')

I get the following

2021-09-15 18:11:46.561864: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-09-15 18:11:46.564618: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcusolver.so.11'; dlerror: libcusolver.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/lib/cuda/include:/usr/lib/cuda/lib64:/usr/local/cuda-11.0/include:/usr/local/cuda-11.0/lib64:/usr/local/cuda/lib64:/usr/local/cuda-11.0/lib64:/usr/lib/cuda/include:/usr/lib/cuda/lib64:
2021-09-15 18:11:46.565065: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1835] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
[]

I have tried a number of other potential setups but nothing works. Any help or suggestions is appreciated.

Here is also a snapshot of nvidia-smi

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.91.03    Driver Version: 460.91.03    CUDA Version: 11.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce GTX 1070    Off  | 00000000:01:00.0  On |                  N/A |
|  0%   53C    P0    34W / 200W |   1175MiB /  8085MiB |      2%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A       934      G   /usr/lib/xorg/Xorg                198MiB |
|    0   N/A  N/A      1542      G   /usr/lib/xorg/Xorg                515MiB |
|    0   N/A  N/A      1674      G   /usr/bin/gnome-shell              227MiB |
|    0   N/A  N/A      2843      G   /usr/lib/firefox/firefox          219MiB |
+-----------------------------------------------------------------------------+
justanewb
  • 111
  • 2
    The Nvidia instructions have you add the PATH and LD_LIBRARY_PATH modifications to your .profile, not .bashrc. In doing that, get rid of the dups, the non-existent dirs (e.g./usr/lib/cuda...) and don't put the ...include in the LD_LIBRARY_PATH. After CUDA works (run the samples), install CDNN and run its samples, then install tensorflow. The tensorflow you select may have specific CUDA and DNN version requirements. – ubfan1 Sep 17 '21 at 04:46
  • @ubfan1 Thank you for your comment. You seem pretty well versed in this, would you mind providing a detailed answer? – justanewb Sep 19 '21 at 19:28
  • Search this site for CUDA. Terrance has a nice answer at https://askubuntu.com/questions/1077061/how-do-i-install-nvidia-and-cuda-drivers-into-ubuntu/1077063#1077063 – ubfan1 Sep 19 '21 at 21:40

0 Answers0