目录
首先,下载 cuda 的 driver,例如440.33.01版本的 driver:
https://www.nvidia.com/download/driverResults.aspx/154570
然后用root安装:
sh NVIDIA-Linux-x86_64-440.33.01.run --no-kernel-module
然后应该就可以nvidia-smi
了,只是出来的可能是:
NVIDIA-SMI 440.33.01 Driver Version: 440.33.01 CUDA Version: N/A
然后,安装 cuda,例如440.33.01版本对应的 cuda10.2:
https://developer.nvidia.com/cuda-10.2-download-archive?target_os=Linux&target_arch=x86_64&target_distro=CentOS&target_version=6&target_type=runfilelocal
也就是:
wget http://developer.download.nvidia.com/compute/cuda/10.2/Prod/local_installers/cuda_10.2.89_440.33.01_rhel6.run --no-check-certificate
然后用root(加tmpdir 是因为默认看的是/tmp,很容易满的):
sh cuda_10.2.89_440.33.01_rhel6.run -tmpdir /home/work/tmp --silent --toolkit
再然后:
[work@xxx.xxx ~]$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:24:38_PDT_2019
Cuda compilation tools, release 10.2, V10.2.89