1. 기존 nvidia 드라이버, cuda 삭제
1-1. sudo apt --purge remove "*cuda*" "*cublas*" "*cufft*" "*cufile*" "*curand*" \
"*cusolver*" "*cusparse*" "*gds-tools*" "*npp*" "*nvjpeg*" "nsight*" "*nvvm*"
1-2. sudo apt --purge remove "*nvidia*" "libxnvctrl*"
1-3. sudo rm /etc/apt/sources.list.d/cuda.list
1-4. sudo rm /etc/apt/sources.list.d/cuda*.list
1-5. sudo apt autoremove
2. apt 업데이트
2-1. sudo apt update
2-2. sudo apt upgrade
3. Cuda 설치
3-1. wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-keyring_1.1-1_all.deb
3-2. sudo dpkg -i cuda-keyring_1.1-1_all.deb
3-3. sudo apt update
3-4. sudo apt install cuda-toolkit-12-4
% 환경 변수 등록
3-5-1. gedit ~/.bashrc
3-5-2. export PATH=/usr/local/cuda/bin:$PATH
3-5-3. export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
3-5-4. source ~/.bashrc
3-6. nvcc --version (설치 확인)
(3-6. 실행 후, cuda가 install 안 됐을 경우만 해당)
3-7. sudo rm /etc/apt/sources.list.d/cuda.list
3-8. sudo wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin -O /etc/apt/preferences.d/cuda-repository-pin-600
3-9. sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub
3-10. sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /"
3-11. sudo apt update
3-12. sudo apt install cuda-toolkit-12-4
4. Nvidia driver 설치
4-1. sudo add-apt-repository ppa:graphics-drivers/ppa
4-2. apt list nvidia-driver-5*
4-3. sudo apt install cuda-drivers-550
or
4-3. sudo apt install nvidia-driver-550 (560 으로 변경 가능)
4-4. nvidia-smi
[참고 자료]
https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=20.04&target_type=deb_network
CUDA Toolkit 12.1 Downloads
Get the latest feature updates to NVIDIA's proprietary compute stack.
developer.nvidia.com
https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html#cuda-major-component-versions__table-cuda-toolkit-driver-versions
1. CUDA 12.6 Release Notes — Release Notes 12.6 documentation
Using long-deprecated cusolverDnPotrf, cusolverDnPotrs, cusolverDnGeqrf, cusolverDnGetrf, cusolverDnGetrs, cusolverDnSyevd, cusolverDnSyevdx, cusolverDnGesvd, and their accompanying bufferSize functions will result in a deprecation warning. The warning can
docs.nvidia.com
https://docs.nvidia.com/deploy/cuda-compatibility/
1. Why CUDA Compatibility — CUDA Compatibility r555 documentation
Branches R525, R515, R510, R465, R460, R455, R450, R440, R418, R410, R396, R390 are end of life and are not supported targets for compatibility.
docs.nvidia.com
'공부 > 프로그래밍' 카테고리의 다른 글
[Julia] Julia에서 파이썬 "사용자 정의 / 로컬" 모듈 사용하기 (2) | 2024.08.22 |
---|---|
[Julia] Julia에서 파이썬 라이브러리 사용하기 (0) | 2024.07.17 |
[Julia] 대용량 데이터 효율적으로 처리하기: MATLAB과 Julia 비교 (0) | 2024.07.11 |
[Julia] 로컬 패키지 만드는 법 (0) | 2024.07.09 |
[zeroMQ] F77_ZMQ Linking using CMake (7) (0) | 2024.07.04 |
댓글