WSL2 CUDA, CUDNN 설정

설치 및 환경 설정

WSL 삭제

wslconfig.exe /l
wslconfig.exe /u linux-version

WSL 설치

Microsoft Store에서 설치

기본값 변경

wsl -s linux-version

부가 설정

python3-pip 설치

sudo apt update -y && sudo apt upgrade -y
apt-get install g++ openjdk-8-jdk python3-dev python3-pip curl -y
python3 -m pip install --upgrade pip
pip install pipenv
sudo apt install pipenv -y

Build from source | TensorFlow

CUDA Toolkit Archive

cuda 버전 맞춰 다운로드

🎱 error시 /var/log/cuda-installer.log 확인 필요

cuda 설치 이후

# /usr/local/cuda/bin/
./nvcc -V

cp -R /usr/local/cuda/bin/* /usr/bin

cudnn 다운로드

cudnn 버전 맞춰 다운로드

tar xvf cudnn-linux-x86_64...

sudo cp include/cudnn* /usr/local/cuda-11.x/include
sudo cp -P lib64/libcudnn* /usr/local/cuda-11.x/lib64
sudo chmod a+r /usr/local/cuda-11.x/include/cudnn* /usr/local/cuda-11.x/lib64/libcudnn*

(cuda-11.x or cuda directory)

sudo nano ~/.bashrc

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.x/lib64/

source ~/.bashrc