Skip to content

Commit

Permalink
update CUDA setup
Browse files Browse the repository at this point in the history
  • Loading branch information
@afrendeiro
afrendeiro committed Mar 18, 2024
1 parent 68fc0af commit b4db8a3
Showing 1 changed file with 54 additions and 84 deletions.
138 changes: 54 additions & 84 deletions _posts/2022-10-02-vm_setup_for_deep_learning.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,23 +10,26 @@ tags: [software, deep learning, ubuntu, pytorch, setup]

# GPU-machine, Ubuntu 22.04

### Low-level stuff

{% highlight bash %}
# Check GPU is CUDA-compatible
lspci | grep -i nvidia

# Remove any previous installations
sudo apt-get purge nvidia*
sudo apt remove nvidia-*
```bash
# Get latest version strings
UBUNTU_VERSION=ubuntu2204
ARCH=amd64
DRIVER_VERSION=$(apt-cache search nvidia-driver- | grep "^nvidia-driver" | grep -v open | sort | tail -n 1 | sed "s/ .*//g" | sed -r 's/.*-([0-9]+)/\1/g')
CUDA_VERSION=$(apt-cache search cuda- | grep "^cuda-[0-9]" | sort | tail -n 1 | sed "s/ .*//g" | sed -r 's/.*-([0-9]+-[0-9])/\1/g')
CUDDN_VERSION=9.0.0

# Clean up old installs
sudo apt-get -y purge nvidia*
sudo apt-get -y remove nvidia-*
sudo rm /etc/apt/sources.list.d/cuda*
sudo apt-get autoremove && sudo apt-get autoclean
sudo apt-get -y autoremove && sudo apt-get -y autoclean
sudo rm -rf /usr/local/cuda*

# Install
# Update
sudo apt-get update
sudo apt-get upgrade -y

# Install base libs
sudo apt-get install -y \
build-essential \
libatlas-base-dev \
Expand All @@ -44,89 +47,56 @@ sudo apt-get install -y \
libx11-dev \
libxmu-dev \
libxi-dev
{% endhighlight %}

### NVIDIA drivers

{% highlight bash %}
# Install NVIDIA drivers
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update
sudo apt install libnvidia-common-470
sudo apt install libnvidia-gl-470
sudo apt install nvidia-driver-470
sudo apt install nvidia-settings
sudo apt install nvidia-utils-470
{% endhighlight %}

### CUDA

{% highlight bash %}

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin

sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600

sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub

sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /"
sudo apt-get update
sudo apt install cuda-11-3


# setup your paths
echo 'export PATH=/usr/local/cuda-11.3/bin:$PATH' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda-11.3/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc
sudo apt-get install -y libnvidia-common-${DRIVER_VERSION}
sudo apt-get install -y libnvidia-gl-${DRIVER_VERSION}
sudo apt-get install -y nvidia-driver-${DRIVER_VERSION}
sudo apt-get install -y nvidia-settings
sudo apt-get install -y nvidia-utils-${DRIVER_VERSION}

# Install CUDA
wget https://developer.download.nvidia.com/compute/cuda/repos/${UBUNTU_VERSION}/x86_64/cuda-${UBUNTU_VERSION}.pin
sudo mv cuda-${UBUNTU_VERSION}.pin /etc/apt/preferences.d/cuda-repository-pin-600
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/${UBUNTU_VERSION}/x86_64/3bf863cc.pub
sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/${UBUNTU_VERSION}/x86_64/ /"
sudo apt-get update
sudo apt-get install -y cuda-${CUDA_VERSION}

# setup CUDA paths (TODO: test if this has already been added)
__export='
# CUDA
if [ -d "/usr/local/cuda/bin/" ]; then
export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/opt/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/opt/cuda/include${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/targets/x86_64-linux/include${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export CUDA_HOME=/usr/local/cuda
fi
'
echo "$__export" >> ~/.bashrc
source ~/.bashrc
sudo ldconfig
{% endhighlight %}

## cuDNN

{% highlight bash %}
# Download .deb for CUDA 11.3 from https://developer.nvidia.com/rdp/cudnn-download, and copy to VM via scp. Then:
# wget https://developer.nvidia.com/compute/cudnn/secure/8.5.0/local_installers/11.7/cudnn-local-repo-ubuntu2204-8.5.0.96_1.0-1_amd64.deb
sudo dpkg - i cudnn-local-repo-ubuntu2204-8.5.0.96_1.0-1_amd64.deb
{% endhighlight %}

## Python and Torch

{% highlight bash %}
# Make sure the `python` command is by default Python3 (it will be 3.10 for Ubuntu 22.04)
sudo apt install python-is-python3

# Install pip
wget https://bootstrap.pypa.io/get-pip.py
python get-pip.py

python -m pip install wheel setuptools
# Install CuDNN: https://developer.nvidia.com/rdp/cudnn-download
wget https://developer.download.nvidia.com/compute/cudnn/${CUDDN_VERSION}/local_installers/cudnn-local-repo-${UBUNTU_VERSION}-${CUDDN_VERSION}_1.0-1_${ARCH}.deb
sudo dpkg -i cudnn-local-repo-${UBUNTU_VERSION}-${CUDDN_VERSION}_1.0-1_${ARCH}.deb
sudo cp /var/cudnn-local-repo-${UBUNTU_VERSION}-${CUDDN_VERSION}/cudnn-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install cudnn

# Install Python scientific stack
python -m pip install \
IPython \
urlpath \
tqdm \
numpy \
scipy \
pandas \
matplotlib \
seaborn \
anndata \
scanpy \
squidpy \
statsmodels \
scikit-learn \
scikit-image \
networkx \
torch \
torchvision \
IPython \
--extra-index-url https://download.pytorch.org/whl/cu113
{% endhighlight %}
# Reboot!
sudo reboot
```


# Non-GPU machine, Ubuntu 22.04

{% highlight bash %}
```bash
# Make sure VM is up-to-date
sudo apt-get update
sudo apt-get upgrade -y
Expand Down Expand Up @@ -184,4 +154,4 @@ python -m pip install \
torchvision \
IPython \
--extra-index-url https://download.pytorch.org/whl/cu113
{% endhighlight %}
```

0 comments on commit b4db8a3

Please sign in to comment.