基于Centos7安装cuda 10

• 环境介绍：
• 一、安装驱动
• 二、安装cuda

环境介绍：

CentOS Linux release 7.9.2009 (Core)

NVIDIA: Tesla K80

一、安装驱动

1、检查显卡版本

lshw -numeric -C display

*-display

       description: 3D controller

       product: GK210GL [Tesla K80] [10DE:102D]

       vendor: NVIDIA Corporation [10DE]

       physical id: 0

       bus info: pci@0000:04:00.0

       logical name: /dev/fb0

       version: a1

       width: 64 bits

       clock: 33MHz

可以看到显卡是Tesla K80的

2、去官方下载驱动

Official Drivers | NVIDIA

选择操作系统及cuda版本

3、安装依赖

yum install kernel-devel gcc -y

4、安装驱动

chmod +x NVIDIA-Linux-x86_64-410.129-diagnostic.run

./NVIDIA-Linux-x86_64-410.129-diagnostic.run

5、报错

报错 unable to find the kernel source tree for the currently running kernel.........，使用下面命令安装，/3.10.0-1160.31.1.el7.x86_64需要改成自己的目录

./NVIDIA-Linux-x86_64-410.129-diagnostic.run --kernel-source-path=/usr/src/kernels/3.10.0-1160.31.1.el7.x86_64 -k $(uname -r)

然后图像化界面操作。基本是一直回车就可以了

6、安装完成后检查

nvidia-smi

Wed Jun 23 20:19:33 2021

+-----------------------------------------------------------------------------+

| NVIDIA-SMI 410.129      Driver Version: 410.129      CUDA Version: 10.0     |

|-------------------------------+----------------------+----------------------+

| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |

| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |

|===============================+======================+======================|

|   0  Tesla K80           Off  | 00000000:04:00.0 Off |                    0 |

| N/A   29C    P0    56W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   1  Tesla K80           Off  | 00000000:05:00.0 Off |                    0 |

| N/A   26C    P0    70W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   2  Tesla K80           Off  | 00000000:08:00.0 Off |                    0 |

| N/A   28C    P0    59W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   3  Tesla K80           Off  | 00000000:09:00.0 Off |                    0 |

| N/A   22C    P0    71W / 149W |      0MiB / 11441MiB |     80%      Default |

+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+

| Processes:                                                       GPU Memory |

|  GPU       PID   Type   Process name                             Usage      |

|=============================================================================|

|  No running processes found                                                 |

+-----------------------------------------------------------------------------+

到这里NVIDIA的驱动就安装完成了。

方法二：使用rpm方式安装

i) `rpm -i nvidia-driver-local-repo-rhel7-418.211.00-1.0-1.x86_64.rpm'
ii) `yum clean all`
iii) `yum install cuda-drivers` 会提示缺少dkms包，手动安装dkms包 dkms-2.3-5.8.noarch.rpm
iv) `reboot`

二、安装cuda

下载链接：CUDA Toolkit Archive | NVIDIA Developer

参考文档：Centos7安装cuda10.1_Happy_wtg的博客-CSDN博客_centos7安装cuda10.1

选择对应的版本进行下载。

检查当前运行级别。

检查当前运行级别, 若为3则不用修改，若为5需要修改为3.

修改运行级别为3的命令：systemctl set-default multi-user.target，重启机器，再次执行runlevel，此时应该变为3.

# runlevel

N 3

# chmod +x cuda_10.0.130_410.48_linux.run

安装

出现这个界面表示安装完成。

# sh cuda_10.0.130_410.48_linux.run --no-opengl-libs

Do you accept the previously read EULA?

accept/decline/quit: accept

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?

(y)es/(n)o/(q)uit: n   #因为上面已经安装过了驱动

Install the CUDA 10.0 Toolkit?

(y)es/(n)o/(q)uit: y

Enter Toolkit Location

 [ default is /usr/local/cuda-10.0 ]:

Do you want to install a symbolic link at /usr/local/cuda?

(y)es/(n)o/(q)uit: y

Install the CUDA 10.0 Samples?

(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location

 [ default is /home/dd ]:

Installing the CUDA Toolkit in /usr/local/cuda-10.0 ...

Missing recommended library: libGLU.so

Missing recommended library: libX11.so

Missing recommended library: libXi.so

Missing recommended library: libXmu.so

Installing the CUDA Samples in /home/dd ...

Copying samples to /home/dd/NVIDIA_CUDA-10.0_Samples now...

Finished copying samples.

===========

= Summary =

===========

Driver:   Not Selected

Toolkit:  Installed in /usr/local/cuda-10.0

Samples:  Installed in /home/dd, but missing recommended libraries

Please make sure that

 -   PATH includes /usr/local/cuda-10.0/bin

 -   LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 384.00 is required for CUDA 10.0 functionality to work.

To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:

    sudo <CudaInstaller>.run -silent -driver

Logfile is /tmp/cuda_install_11398.log

查看运行状态：nvidia-smi

Wed Jun 23 22:00:33 2021

+-----------------------------------------------------------------------------+

| NVIDIA-SMI 410.129      Driver Version: 410.129      CUDA Version: 10.0     |

|-------------------------------+----------------------+----------------------+

| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |

| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |

|===============================+======================+======================|

|   0  Tesla K80           Off  | 00000000:04:00.0 Off |                    0 |

| N/A   31C    P0    56W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   1  Tesla K80           Off  | 00000000:05:00.0 Off |                    0 |

| N/A   27C    P0    70W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   2  Tesla K80           Off  | 00000000:08:00.0 Off |                    0 |

| N/A   31C    P0    60W / 149W |      0MiB / 11441MiB |      0%      Default |

+-------------------------------+----------------------+----------------------+

|   3  Tesla K80           Off  | 00000000:09:00.0 Off |                    0 |

| N/A   24C    P0    71W / 149W |      0MiB / 11441MiB |     78%      Default |

+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+

| Processes:                                                       GPU Memory |

|  GPU       PID   Type   Process name                             Usage      |

|=============================================================================|

|  No running processes found                                                 |

+-----------------------------------------------------------------------------+

配置环境变量

# vim /etc/profile

......

#cuda

export PATH=/usr/local/cuda/bin:$PATH

export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

cuda测试

测试 cuda, nvcc 命令是否可用

# cuda ; 按两下 tab 键

cudafe++                      cuda-gdb                      cuda-gdbserver                cuda-install-samples-10.0.sh  cuda-memcheck

# nvcc --version

nvcc: NVIDIA (R) Cuda compiler driver

Copyright (c) 2005-2018 NVIDIA Corporation

Built on Sat_Aug_25_21:08:01_CDT_2018

Cuda compilation tools, release 10.0, V10.0.130

用 cuda 例程测试，找到例程的安装目录

目录可以查看上面cuda安装完成提示那里。

默认在 /root 下。

我这里用的dd这个用户安装的所以是在/home/dd/NVIDIA_CUDA-10.0_Samples

cd /home/dd/NVIDIA_CUDA-10.0_Samples

# ls

0_Simple  1_Utilities  2_Graphics  3_Imaging  4_Finance  5_Simulations  6_Advanced  7_CUDALibraries  bin  common  EULA.txt  Makefile

只需要挑选其中的几个进行测试即可，比如

# cd 1_Utilities/deviceQuery

# make

# ./deviceQuery

./deviceQuery Starting...

CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 4 CUDA Capable device(s)

Device 0: "Tesla K80"

  CUDA Driver Version / Runtime Version          10.0 / 10.0

  CUDA Capability Major/Minor version number:    3.7

  Total amount of global memory:                 11441 MBytes (11996954624 bytes)

  (13) Multiprocessors, (192) CUDA Cores/MP:     2496 CUDA Cores

  GPU Max Clock rate:                            824 MHz (0.82 GHz)

  Memory Clock rate:                             2505 Mhz

  Memory Bus Width:                              384-bit

  L2 Cache Size:                                 1572864 bytes

  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)

  Maximum Layered 1D Texture Size, (num) layers  1D=(16384), 2048 layers

  Maximum Layered 2D Texture Size, (num) layers  2D=(16384, 16384), 2048 layers

  Total amount of constant memory:               65536 bytes

  Total amount of shared memory per block:       49152 bytes

  Total number of registers available per block: 65536

  Warp size:                                     32

至此，CUDA Toolkit 已经安装完成。