GeForce 710M 双显卡笔记本安装 CentOS 7 + CUDA 9.1 踩坑手记

 

目录

 

几个简单的问题与答案

简单的故事背景

安装 CentOS 7

制作启动U盘

安装必定会遇上的坑

安装编译环境

删掉 nouveau

修改 blacklist.conf

修改grub

重新生成 initramfs image file

确认nouveau被成功禁用

安装 Nvidia 驱动

检测需要安装哪个版本的驱动

下载驱动

安装驱动

验证

安装 CUDA

下载 CUDA

安装CUDA

测试CUDA

使用 nvcc 来初步测试

编译 CUDA 的示例程序来进行测试

deviceQuery

bandwidthTest

打补丁

---

几个简单的问题与答案

1. GeForce 710M 支持CUDA吗 ？ 支持
2. GeForce 710M 对应的显卡驱动版本是什么？390.xx
3. GeForce 710M 支持的CUDA版本是什么？CUDA 9.1

简单的故事背景

这些年的MacBook Pro 的独显都是A卡，想玩点机器学习，又不想花钱买新卡，遂翻出了七年前的一台双显卡的老本，折腾起来，玩玩物体识别。期间碰到了很多问题，逐一解决之后，把它们都记录了下来。

老本 (Acer 超级本) 的配置：

• Intel(R) Core(TM) i5-3337U CPU @ 1.80GHz

• Nvidia GeForce 710M

• 8G 内存

• 120G SSD

安装好 Nvidia 驱动后，执行 nvidia-smi 结果如下

[root@Acer-CentOS7 deviceQuery]# nvidia-smi 
Sun Oct 25 22:05:00 2020       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.138                Driver Version: 390.138                   |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce 710M        Off  | 00000000:01:00.0 N/A |                  N/A |
| N/A   55C    P0    N/A /  N/A |      0MiB /  1985MiB |     N/A      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0                    Not Supported                                       |
+-----------------------------------------------------------------------------+

安装好后编译执行 cuda 示例 deviceQuery ，结果如下

[root@Acer-CentOS7 deviceQuery]# ./deviceQuery 
./deviceQuery Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "GeForce 710M"
  CUDA Driver Version / Runtime Version          9.1 / 9.1
  CUDA Capability Major/Minor version number:    2.1
  Total amount of global memory:                 1985 MBytes (2081816576 bytes)
MapSMtoCores for SM 2.1 is undefined.  Default to use 64 Cores/SM
MapSMtoCores for SM 2.1 is undefined.  Default to use 64 Cores/SM
  ( 2) Multiprocessors, ( 64) CUDA Cores/MP:     128 CUDA Cores
  GPU Max Clock rate:                            1550 MHz (1.55 GHz)
  Memory Clock rate:                             900 Mhz
  Memory Bus Width:                              64-bit
  L2 Cache Size:                                 131072 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65535), 3D=(2048, 2048, 2048)
  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: 32768
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (65535, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Supports Cooperative Kernel Launch:            No
  Supports MultiDevice Co-op Kernel Launch:      No
  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 9.1, CUDA Runtime Version = 9.1, NumDevs = 1
Result = PASS

安装 CentOS 7

制作启动U盘

我下载的是 Everything DVD 版本，参考链接 http://mirrors.163.com/centos/7/isos/x86_64/CentOS-7-x86_64-Everything-2003.iso

下载后使用UltraISO软碟通 写入硬盘镜像即可。

安装必定会遇上的坑

Warning: /dev/root does not exist 

参考解决办法：https://blog.csdn.net/bajiudongfeng/article/details/47732377

根据搜到的资料说到，由于 磁盘 Label的长度有限，原始磁盘Label 应为 CentOS 7 x86_64,最终被截断了，变为了 CentOS 7 x8，所以我的解决办法是修改启动菜单时把这个 Label 改一下再启动。具体操作如下：

1. 进入到第一个界面时，上下键移动到 Install CentOS 7 这个菜单上
2. 按 e 健进入编辑状态
3. 将 vmlinuz initrd=initrd.imginst.stage2=hd:LABEL=CentOS\x207\x20x86_64 rd.live.check quiet 改为 vmlinuz initrd=initrd.imginst.stage2=hd:LABEL=CentOS\x207\x20x8 rd.live.check quiet
4. (上一步就是去掉 6_64)
5. 按 Ctrl + x 启动

安装编译环境

# 安装开发工具
yum groupinstall "Development Tools"

# 安装内核开发库和头文件
yum install kernel-devel kernel-headers



# 上一步安装的东东 和 内核版本 可能存在不一致的情况，执行以下操作即可同步

yum distro-sync

# 查看内核
[root@Acer-CentOS7 bandwidthTest]# uname -r
3.10.0-1127.19.1.el7.x86_64

# 查看安装的 kernel-devel kernel-headers 版本

[root@Acer-CentOS7 bandwidthTest]# rpm -qa |grep kernel
kernel-devel-3.10.0-1127.19.1.el7.x86_64
kernel-3.10.0-1062.el7.x86_64
kernel-headers-3.10.0-1127.19.1.el7.x86_64
kernel-3.10.0-1127.19.1.el7.x86_64
kernel-tools-libs-3.10.0-1127.19.1.el7.x86_64
kernel-tools-3.10.0-1127.19.1.el7.x86_64
kernel-doc-3.10.0-1127.19.1.el7.noarch
abrt-addon-kerneloops-2.1.11-57.el7.centos.x86_64

删掉 nouveau

修改 blacklist.conf

这是一个第三方开源的Nvidia 驱动，必须禁用掉，才能装 Nvidia 的驱动。

# 找到这一行，在行前加井号，注释掉
blacklist nvidiafb
# 增加以下两行
blacklist nouveau
options nouveau modeset=0

[root@Acer-CentOS7 deviceQuery]# vim /lib/modprobe.d/dist-blacklist.conf 

#
# Listing a module here prevents the hotplug scripts from loading it.
# Usually that'd be so that some other driver will bind it instead,
# no matter which driver happens to get probed first.  Sometimes user
# mode tools can also control driver binding.
#
# Syntax: see modprobe.conf(5).
#

# watchdog drivers
blacklist i8xx_tco

# framebuffer drivers
blacklist aty128fb
blacklist atyfb
blacklist radeonfb
blacklist i810fb
blacklist cirrusfb
blacklist intelfb
blacklist kyrofb
blacklist i2c-matroxfb
blacklist hgafb
# 找到这一行，注释掉
#blacklist nvidiafb
# 增加以下两行
blacklist nouveau
options nouveau modeset=0
blacklist rivafb
blacklist savagefb
blacklist sstfb
blacklist neofb
blacklist tridentfb
blacklist tdfxfb
blacklist virgefb
blacklist vga16fb
blacklist viafb

# ISDN - see bugs 154799, 159068
blacklist hisax
blacklist hisax_fcpcipnp

# sound drivers
blacklist snd-pcsp

# I/O dynamic configuration support for s390x (bz #563228)
blacklist chsc_sch

# crypto algorithms
blacklist sha1-mb

# see bz #1562114
blacklist sha256-mb
blacklist sha512-mb

修改grub

# 修改  /etc/default/grub
# 在 GRUB_CMDLINE_LINUX 行尾部、双引号内 后增加内容 rdblacklist=nouveau
# 文件内容如下：

GRUB_TIMEOUT=5
GRUB_DISTRIBUTOR="$(sed 's, release .*$,,g' /etc/system-release)"
GRUB_DEFAULT=saved
GRUB_DISABLE_SUBMENU=true
GRUB_TERMINAL_OUTPUT="console"
GRUB_CMDLINE_LINUX="crashkernel=auto rd.lvm.lv=centos/root rd.lvm.lv=centos/swap rhgb quiet rdblacklist=nouveau"
GRUB_DISABLE_RECOVERY="true"

重新生成配置

# 重新生成grub.cfg

grub2-mkconfig -o /boot/grub2/grub.cfg


# 如果是EFI启动，不是BIOS的话，应该是 ，自己判断下路径，find 也行呢

grub2-mkconfig -o /boot/efi/EFI/centos/grub.cfg

重新生成 initramfs image file

这一步非常重要，要不然无法禁用 nouveau

# 进入 /boot/
[root@Acer-CentOS7 ~]# cd /boot/

# 找到当前使用的内核版本所对应的 initramfs
[root@Acer-CentOS7 boot]# ll
总用量 283840
-rw-r--r--. 1 root root   152976 8月   7 2019 config-3.10.0-1062.el7.x86_64
-rw-r--r--. 1 root root   153567 8月  25 13:27 config-3.10.0-1127.19.1.el7.x86_64
drwx------. 3 root root    16384 12月 31 1969 efi
drwxr-xr-x. 2 root root       27 10月 25 12:41 grub
drwx------. 2 root root       37 10月 25 14:07 grub2
-rw-------. 1 root root 78623661 10月 25 12:47 initramfs-0-rescue-9499a8e5ead6495493bc04bb41e58279.img
-rw-------. 1 root root 42354323 10月 25 15:13 initramfs-3.10.0-1062.el7.x86_64.img
-rw-------. 1 root root 33766003 10月 25 13:50 initramfs-3.10.0-1062.el7.x86_64.img.bak
-rw-------. 1 root root 15670688 10月 25 14:05 initramfs-3.10.0-1062.el7.x86_64kdump.img
-rw-------. 1 root root 42362518 10月 25 15:14 initramfs-3.10.0-1127.19.1.el7.x86_64.img
-rw-------. 1 root root 33759904 10月 25 14:13 initramfs-3.10.0-1127.19.1.el7.x86_64.img.bak
-rw-------. 1 root root 15675591 10月 25 14:11 initramfs-3.10.0-1127.19.1.el7.x86_64kdump.img
-rw-r--r--. 1 root root   318717 8月   7 2019 symvers-3.10.0-1062.el7.x86_64.gz
-rw-r--r--. 1 root root   320536 8月  25 13:27 symvers-3.10.0-1127.19.1.el7.x86_64.gz
-rw-------. 1 root root  3594971 8月   7 2019 System.map-3.10.0-1062.el7.x86_64
-rw-------. 1 root root  3612420 8月  25 13:27 System.map-3.10.0-1127.19.1.el7.x86_64
-rwxr-xr-x. 1 root root  6734016 10月 25 12:47 vmlinuz-0-rescue-9499a8e5ead6495493bc04bb41e58279
-rwxr-xr-x. 1 root root  6734016 8月   7 2019 vmlinuz-3.10.0-1062.el7.x86_64
-rwxr-xr-x. 1 root root  6765160 8月  25 13:27 vmlinuz-3.10.0-1127.19.1.el7.x86_64


# 我这里是 1127
# 备份是个好习惯
cp initramfs-3.10.0-1127.19.1.el7.x86_64.img initramfs-3.10.0-1127.19.1.el7.x86_64.img.bak

# 执行 dracut
dracut -v initramfs-3.10.0-1127.19.1.el7.x86_64.img --force

# 等待成功执行之后，重启机器

reboot

确认nouveau被成功禁用

# 执行 lsmod | grep nouveau 没有任何输出，则已经成功禁用
# 如果还有其他输出，请重来以上步骤吧
[root@Acer-CentOS7 boot]# lsmod | grep nouveau
[root@Acer-CentOS7 boot]# 

安装 Nvidia 驱动

检测需要安装哪个版本的驱动

检测显卡

[root@Acer-CentOS7 boot]# lspci | grep -i "vga\|nvidia"
00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)
01:00.0 3D controller: NVIDIA Corporation GF117M [GeForce 610M/710M/810M/820M / GT 620M/625M/630M/720M] (rev a1)


# 如果找不到命令，安装一下
# yum install pciutils

检测显卡需要的NVIDIA驱动版本

# 这一步操作参考 http://elrepo.org/tiki/HomePage

[root@Acer-CentOS7 ~]# rpm --import https://www.elrepo.org/RPM-GPG-KEY-elrepo.org
[root@Acer-CentOS7 ~]# yum install https://www.elrepo.org/elrepo-release-7.el7.elrepo.noarch.rpm
已加载插件：fastestmirror, langpacks
elrepo-release-7.el7.elrepo.noarch.rpm                                                                                                                                                                 | 8.6 kB  00:00:00     
正在检查 /var/tmp/yum-root-phlquT/elrepo-release-7.el7.elrepo.noarch.rpm: elrepo-release-7.0-5.el7.elrepo.noarch
/var/tmp/yum-root-phlquT/elrepo-release-7.el7.elrepo.noarch.rpm 将被安装
正在解决依赖关系
--> 正在检查事务
---> 软件包 elrepo-release.noarch.0.7.0-5.el7.elrepo 将被 安装
--> 解决依赖关系完成

依赖关系解决

==============================================================================================================================================================================================================================
 Package                                           架构                                      版本                                                源                                                                      大小
==============================================================================================================================================================================================================================
正在安装:
 elrepo-release                                    noarch                                    7.0-5.el7.elrepo                                    /elrepo-release-7.el7.elrepo.noarch                                    5.0 k

事务概要
==============================================================================================================================================================================================================================
安装  1 软件包

总计：5.0 k
安装大小：5.0 k
Is this ok [y/d/N]: y
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  正在安装    : elrepo-release-7.0-5.el7.elrepo.noarch                                                                                                                                                                    1/1 
  验证中      : elrepo-release-7.0-5.el7.elrepo.noarch                                                                                                                                                                    1/1 

已安装:
  elrepo-release.noarch 0:7.0-5.el7.elrepo                                                                                                                                                                                    

完毕！
[root@Acer-CentOS7 ~]# yum -y install nvidia-detect
已加载插件：fastestmirror, langpacks
Loading mirror speeds from cached hostfile
 * elrepo: mirrors.tuna.tsinghua.edu.cn
elrepo                                                                                                                                                                                                 | 2.9 kB  00:00:00     
elrepo/primary_db                                                                                                                                                                                      | 479 kB  00:00:01     
正在解决依赖关系
--> 正在检查事务
---> 软件包 nvidia-detect.x86_64.0.440.64-1.el7.elrepo 将被 安装
--> 解决依赖关系完成

依赖关系解决

==============================================================================================================================================================================================================================
 Package                                                架构                                            版本                                                            源                                               大小
==============================================================================================================================================================================================================================
正在安装:
 nvidia-detect                                          x86_64                                          440.64-1.el7.elrepo                                             elrepo                                           25 k

事务概要
==============================================================================================================================================================================================================================
安装  1 软件包

总下载量：25 k
安装大小：37 k
Downloading packages:
nvidia-detect-440.64-1.el7.elrepo.x86_64.rpm                                                                                                                                                           |  25 kB  00:00:00     
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  正在安装    : nvidia-detect-440.64-1.el7.elrepo.x86_64                                                                                                                                                                  1/1 
  验证中      : nvidia-detect-440.64-1.el7.elrepo.x86_64                                                                                                                                                                  1/1 

已安装:
  nvidia-detect.x86_64 0:440.64-1.el7.elrepo                                                                                                                                                                                  

完毕！


[root@Acer-CentOS7 ~]# nvidia-detect -v
Probing for supported NVIDIA devices...
[10de:1140] NVIDIA Corporation GF117M [GeForce 610M/710M/810M/820M / GT 620M/625M/630M/720M]
This device requires the legacy 390.xx NVIDIA driver kmod-nvidia-390xx
[8086:0166] Intel Corporation 3rd Gen Core processor Graphics Controller
An Intel display controller was also detected

使用 nvidia-detect 检测得到的结果为 390.xx

下载驱动

打开 https://www.nvidia.com/en-us/drivers/unix/ 这个网站，下载对应版本的驱动即可！如下图

点击即可下载对应版本的驱动

 

根据提示，会到达这个页面，其实可以使用下载工具下载

可以看到下载链接为 https://us.download.nvidia.com/XFree86/Linux-x86_64/390.138/NVIDIA-Linux-x86_64-390.138.run

安装驱动

下载下来的 *.run 其实是可执行的，按提示安装即可

sh NVIDIA-Linux-x86_64-390.138.run

没有安装桌面环境的，以上操作即可，但已经安装桌面环境的系统会报异常，这时候需要多加一个参数进行安装。

sh NVIDIA-Linux-x86_64-390.138.run -no-x-check



[root@Acer-CentOS7 ~]# sh NVIDIA-Linux-x86_64-390.138.run  -no-x-check
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 390.138...............

验证

安装完成后，使用 nvidia-smi 来检测显卡信息

[root@Acer-CentOS7 deviceQuery]# nvidia-smi
Mon Oct 26 05:07:24 2020       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.138                Driver Version: 390.138                   |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce 710M        Off  | 00000000:01:00.0 N/A |                  N/A |
| N/A   55C    P0    N/A /  N/A |      0MiB /  1985MiB |     N/A      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0                    Not Supported                                       |
+-----------------------------------------------------------------------------+

安装 CUDA

下载 CUDA

cuda 支持的版本和对应显卡驱动的关系在 https://docs.nvidia.com/deploy/cuda-compatibility/index.html 这里有列，简单的说明一下怎么看这个关系。

所以我们要安装的 CUDA 版本为 9.1 

打开 CUDA 下载页面 https://developer.nvidia.com/cuda-toolkit-archive  ，选择 9.1

 

点击后进入到如下的下载页面，依次选择

• Linux 
• x86_64
• CentOS
• 7
• runfile (local)

可以看到有一个安装包 1.6GB，另外有 3个补丁包，都把它们下载了。

安装CUDA

# 执行安装
sh cuda_9.1.85_387.26_linux


# 会展示一堆用户许可协议，可以直接按 q ，然后按提示操作即可，以下是部分交互

# 输入 accept 回车 同意 NVIDIA Software License Agreement

1. NVIDIA Software License Agreement
Do you accept the previously read EULA?
accept/decline/quit: accept

# 是否安装CUDA安装包中的驱动，驱动版本是 387.26 ? 明显不必安装，我们都已经安装了390.138 了...

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 387.26?
(y)es/(n)o/(q)uit: n

# 肯定 y 了
Install the CUDA 9.1 Toolkit?
(y)es/(n)o/(q)uit: y

# 默认，直接回车
Enter Toolkit Location
 [ default is /usr/local/cuda-9.1 ]: 

Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y

Install the CUDA 9.1 Samples?
(y)es/(n)o/(q)uit: y

# 这里它会释放一个文件及，里面包含很多CUDA示例，等会我们测试也要用到
Enter CUDA Samples Location
 [ default is /root ]: 

Installing the CUDA Toolkit in /usr/local/cuda-9.1 ...
Installing the CUDA Samples in /root ...
Copying samples to /root/NVIDIA_CUDA-9.1_Samples now...
Finished copying samples.

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-9.1
Samples:  Installed in /root

Please make sure that
 -   PATH includes /usr/local/cuda-9.1/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-9.1/lib64, or, add /usr/local/cuda-9.1/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-9.1/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-9.1/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 9.1 functionality to work.
To install the driver using this installer, run the following command, replacing  with the name of this run file:
    sudo .run -silent -driver

Logfile is /tmp/cuda_install_8091.log

根据安装提示，需要把 /usr/local/cuda-9.1/bin 加入 PATH 环境变量，把  /usr/local/cuda-9.1/lib64 加入到 LD_LIBRARY_PATH

根据自己的shell来修改，我这儿用的是bash，所以编辑 ~/.bash_profile ,增加两行

PATH=$PATH:$HOME/bin:/usr/local/cuda-9.1/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-9.1/lib64


#source 一下，或者重新登录

source ~/.bash_profile

测试CUDA

使用 nvcc 来初步测试

[root@Acer-CentOS7 deviceQuery]# nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Fri_Nov__3_21:07:56_CDT_2017
Cuda compilation tools, release 9.1, V9.1.85

编译 CUDA 的示例程序来进行测试

deviceQuery

# 我安装时示例代码放到 /root/ 下
[root@Acer-CentOS7 deviceQuery]# cd /root/NVIDIA_CUDA-9.1_Samples/1_Utilities/deviceQuery
[root@Acer-CentOS7 deviceQuery]# make
"/usr/local/cuda-9.1"/bin/nvcc -ccbin g++ -I../../common/inc  -m64    -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_70,code=compute_70 -o deviceQuery.o -c deviceQuery.cpp
"/usr/local/cuda-9.1"/bin/nvcc -ccbin g++   -m64      -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_70,code=compute_70 -o deviceQuery deviceQuery.o 
mkdir -p ../../bin/x86_64/linux/release
cp deviceQuery ../../bin/x86_64/linux/release
[root@Acer-CentOS7 deviceQuery]# ./deviceQuery 
./deviceQuery Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "GeForce 710M"
  CUDA Driver Version / Runtime Version          9.1 / 9.1
  CUDA Capability Major/Minor version number:    2.1
  Total amount of global memory:                 1985 MBytes (2081816576 bytes)
MapSMtoCores for SM 2.1 is undefined.  Default to use 64 Cores/SM
MapSMtoCores for SM 2.1 is undefined.  Default to use 64 Cores/SM
  ( 2) Multiprocessors, ( 64) CUDA Cores/MP:     128 CUDA Cores
  GPU Max Clock rate:                            1550 MHz (1.55 GHz)
  Memory Clock rate:                             900 Mhz
  Memory Bus Width:                              64-bit
  L2 Cache Size:                                 131072 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65535), 3D=(2048, 2048, 2048)
  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: 32768
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (65535, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Supports Cooperative Kernel Launch:            No
  Supports MultiDevice Co-op Kernel Launch:      No
  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 9.1, CUDA Runtime Version = 9.1, NumDevs = 1
Result = PASS

bandwidthTest

[root@Acer-CentOS7 bandwidthTest]# cd /root/NVIDIA_CUDA-9.1_Samples/1_Utilities/bandwidthTest
[root@Acer-CentOS7 bandwidthTest]# make
"/usr/local/cuda-9.1"/bin/nvcc -ccbin g++ -I../../common/inc  -m64    -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_70,code=compute_70 -o bandwidthTest.o -c bandwidthTest.cu
"/usr/local/cuda-9.1"/bin/nvcc -ccbin g++   -m64      -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_70,code=compute_70 -o bandwidthTest bandwidthTest.o 
mkdir -p ../../bin/x86_64/linux/release
cp bandwidthTest ../../bin/x86_64/linux/release
[root@Acer-CentOS7 bandwidthTest]# ./bandwidthTest 
[CUDA Bandwidth Test] - Starting...
Running on...

 Device 0: GeForce 710M
 Quick Mode

 Host to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(MB/s)
   33554432			2843.6

 Device to Host Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(MB/s)
   33554432			3291.3

 Device to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(MB/s)
   33554432			12519.4

Result = PASS

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.

打补丁

打补丁也没啥好记录的，依次执行1-3即可。

sh cuda_9.1.85.1_linux
sh cuda_9.1.85.2_linux
sh cuda_9.1.85.3_linux