深度学习-第一章-Window下CUDA和CUDNN安装教程

• 一、前言
• 二、官方教程
• 三、版本选择
• 四、下载CUDA
• 五、安装CUDA
• • 1、选择解压路径
  • 2、解压完之后系统检查
  • 3、选择自定义安装
  • 4、测试安装是否成功
• 六、下载cuDNN
• 七、安装
• • 1、解压缩
  • 2、添加Path环境变量-系统变量
  • 3、验证安装是否成功
  • • 执行bandwidthTest.exe
    • 执行deviceQuery.exe

windows11 版本安装 CUDA ，首先需要下载两个安装包

• CUDA toolkit（toolkit就是指工具包）
• cuDNN

注：cuDNN 是用于配置深度学习使用

CUDA：Installation Guide Windows :: CUDA Toolkit Documentation

cuDNN：Installation Guide :: NVIDIA Deep Learning cuDNN Documentation

查看支持CUDA的GPU显卡型号：https://developer.nvidia.com/cuda-gpus

CUDA的版本选择需要根据电脑GPU的型号来决定，比如是nvidia卡
在电脑桌面，点击右键菜单找到NVIDIA控制面板打开如下页面：

点击左下角的系统信息，然后在弹框中切换到组件tab

可以发现我们电脑支持的版本是10.2

各个版本下载地址可以参考：https://blog.csdn.net/weixin_44177494/article/details/120444922

最新版下载页面：https://developer.nvidia.com/cuda-downloads
10.2下载页面：https://developer.nvidia.com/cuda-10.2-download-archive

10.2下载地址：https://developer.download.nvidia.cn/compute/cuda/10.2/Prod/local_installers/cuda_10.2.89_441.22_win10.exe

可以在其他盘新建一个临时文件夹，用来存储解压之后的内容，也可以默认

第一次使用选择自定义安装

可以选择默认，也可以选择自定义位置，提前在文件夹创建好

D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2
D:cudaNVIDIA CorporationCUDA Samplesv10.2

运行cmd，输入nvcc --version或者nvcc -V,即可查看版本号

查看安装路径

set cuda

cuDNN地址如下，不过要注意的是，我们需要注册一个账号，才可以进入到下载界面。大家可以放心注册的。
https://developer.nvidia.com/rdp/cudnn-download

可以使用下面网址，查看适配的 cuDNN

cuDNN Archive | NVIDIA Developer

选择跟自己的cuda版本适配的cudnn版本

下载地址：Download cuDNN v8.3.3 (March 18th, 2022), for CUDA 10.2

• 先找到cuda的安装路径，比如我们是：D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2
• 然后把cudnn解压后的三个文件夹复制进入上面cuda的目录
• 由于bin目录等已经存在，则会合并现在大家应该可以理解，cuDNN 其实就是 CUDA 的一个补丁而已，专为深度学习运算进行优化的。然后再参加环境变量

D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2bin
D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2include
D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2lib
D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2libnvvp

配置完成后，我们可以验证是否配置成功，主要使用CUDA内置的deviceQuery.exe 和 bandwithTest.exe：

首先win+R启动cmd，cd到安装目录下的 …extrasdemo_suite,然后分别执行bandwidthTest.exe和deviceQuery.exe,应该得到下图:

D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2extrasdemo_suite>bandwidthTest.exe
[CUDA Bandwidth Test] - Starting...
Running on...

 Device 0: GeForce MX250
 Quick Mode

 Host to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(MB/s)
   33554432                     3043.2

 Device to Host Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(MB/s)
   33554432                     3194.2

 Device to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(MB/s)
   33554432                     47408.0

Result = PASS

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

D:cudaNVIDIA GPU Computing ToolkitCUDAv10.2extrasdemo_suite>deviceQuery.exe
deviceQuery.exe Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "GeForce MX250"
  CUDA Driver Version / Runtime Version          10.2 / 10.2
  CUDA Capability Major/Minor version number:    6.1
  Total amount of global memory:                 2048 MBytes (2147483648 bytes)
  ( 3) Multiprocessors, (128) CUDA Cores/MP:     384 CUDA Cores
  GPU Max Clock rate:                            1582 MHz (1.58 GHz)
  Memory Clock rate:                             3504 Mhz
  Memory Bus Width:                              64-bit
  L2 Cache Size:                                 524288 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               zu bytes
  Total amount of shared memory per block:       zu bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  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): (2147483647, 65535, 65535)
  Maximum memory pitch:                          zu bytes
  Texture alignment:                             zu bytes
  Concurrent copy and kernel execution:          Yes with 5 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  CUDA Device Driver Mode (TCC or WDDM):         WDDM (Windows Display Driver Model)
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            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 = 10.2, CUDA Runtime Version = 10.2, NumDevs = 1, Device0 = GeForce MX250
Result = PASS