Goal
I would like to run a muscle simulation demo found here: GitHub - vcg-uvic/viper: Open-source code for VIPER -- Volume Invariant Position-based Elastic Rods
One of their instructions is to set export LD_LIBRARY_PATH=/usr/local/cuda/lib64
.
However, there is no cuda
folder in local
.
I would like to know to what path I’ve to set LD_LIBRARY_PATH
.
Should be installed
I have been using CUDA for deep learning, installed indirectly when installing PyTorch through to Anaconda Python package manager.
I installed it with the following command: conda install pytorch torchvision cudatoolkit=10.2 -c pytorch
.
Also, I can use GPU accelerated rendering in Blender.
Trying to find CUDA
This means CUDA has to be somewhere on my Ubuntu 18.04 system. Running the following commands returns various paths and information:
$ locate cuda | grep /cuda$
/home/user/Downloads/blender-2.83.3-linux64/2.83/scripts/addons/cycles/source/kernel/kernels/cuda
...
/home/user/anaconda3/lib/python3.7/site-packages/torch/cuda
...
/home/user/anaconda3/pkgs/pytorch-1.6.0-py3.7_cuda10.2.89_cudnn7.6.5_0/lib/python3.7/site-packages/torch/cuda
...
However, adding /lib64
to the command ( locate cuda | grep /cuda/lib64$
) returns nothing.
Some more output:
$ nvidia-smi
Sat Sep 5 18:01:44 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.66 Driver Version: 450.66 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| 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 108... Off | 00000000:26:00.0 On | N/A |
| 33% 43C P0 67W / 260W | 1092MiB / 11176MiB | 5% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 9015 G /usr/lib/xorg/Xorg 737MiB |
| 0 N/A N/A 9145 G /usr/bin/gnome-shell 343MiB |
| 0 N/A N/A 11901 G /usr/lib/firefox/firefox 2MiB |
| 0 N/A N/A 11987 G /usr/lib/firefox/firefox 5MiB |
+-----------------------------------------------------------------------------+
Installing CUDA Toolkit from the developer website - attempt
Also, after downloading CUDA Toolkit 10.2
runfile from the developer website, and executing the command: sudo sh cuda_10.2.89_440.33.01_linux.run
,
I get the following warning:
Existing package manager installation of the driver found. It is strongly recommended that you remove this before continuing.
Question
To what folder do I have to set LD_LIBRARY_PATH
? / How do I find the location of CUDA, or more specifically: cuda/lib64
?