win7 CUDA5 VS2012 安装

我的显卡

参考官方页面http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-microsoft-windows/index.html

下载cuda5：https://developer.nvidia.com/cuda-downloads

The CUDA 5 Installers include the CUDA Toolkit, SDK code samples, and developer drivers.

从这里我们知道只要下载一个包安装就行了

然后更新系统环境变量

CUDA_PATH_V5_0=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\
CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\
CUDA_BIN_PATH=%CUDA_PATH%bin
CUDA_INC_PATH=%CUDA_PATH%include
CUDA_LIB_PATH=%CUDA_PATH%lib\Win32
CUDA_SDK_PARH=C:\ProgramData\NVIDIA Corporation\CUDA Samples\v5.0\
CUDA_SDK_LIB_PARH=%CUDA_SDK_PARH%common\lib\Win32
CUDA_SDK_BIN_PARH=%CUDA_SDK_PARH%bin\win32

在Path里，加这些

;%CUDA_PATH%;%CUDA_SDK_PARH%;%CUDA_LIB_PATH%;%CUDA_BIN_PATH%;%CUDA_SDK_LIB_PARH%;%CUDA_SDK_BIN_PARH%

复制

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\extras\visual_studio_integration\MSBuildExtensions

文件夹下的文件到

C:\Program Files\MSBuild\Microsoft.Cpp\v4.0\V110\BuildCustomizations\MSBuildExtensions

让CUDA5和VS2012兼容，还需要做如下修改

打开CUDA 5.0.props

在

<CudaClVersion Condition="'$(PlatformToolset)' == 'v90'">2008</CudaClVersion>
<CudaClVersion Condition="'$(PlatformToolset)' == 'v100'">2010</CudaClVersion>

下面添加

<CudaClVersion Condition="'$(PlatformToolset)' == 'v110'">2010</CudaClVersion>

打开 CUDA 5.0.targets

找到“CudaCleanDependsOn”

将标签里的东西修改成

<CudaCleanDependsOn>
  $(CudaCompileDependsOn);
  _SelectedFiles;
  CudaFilterSelectedFiles;
  AddCudaCompileMetadata;
  AddCudaLinkMetadata;
  AddCudaCompileDeps;
  AddCudaCompilePropsDeps;
  ValidateCudaBuild;
  ValidateCudaCodeGeneration;
  ComputeCudaCompileOutput;
  PrepareForCudaBuild
</CudaCleanDependsOn>

同样在 CUDA 5.0.targets里搜索“GenerateRelocatableDeviceCode”，修改成

GenerateRelocatableDeviceCode="%(CudaCompile.GenerateRelocatableDeviceCode)"

在该文件下面一点的地方查找 “CodeGeneration”，修改为

CodeGeneration="%(CudaCompile.CodeGenerationValues)"

同样是这个文件搜索 “CommandLineTemplate”，一般在最后，修改为(注：官方整了个html代码，中文很多地方都是copy，等号后面应该像这样才ok。。。)

CommandLineTemplate="(CudaToolkitNvccPath) %(CudaCompile.BuildCommandLineTemplate) %(CudaCompile.ApiCommandLineTemplate) %(CudaCompile.CleanCommandLineTemplate)" />

编辑

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\include\host_config.h

修改

#if _MSC_VER < 1400 || _MSC_VER > 1600

为

#if _MSC_VER < 1400 || _MSC_VER > 1700

配置VS2012，依路径“Tools|Options|Text Editor|File Extension（工具|选项|文本编辑器|文件扩展名）”，下拉框选择Microsoft Visual C++，扩展名添加cu和cuh

重启VS

设置CUDA关键词高亮

复制

C:\ProgramData\NVIDIA Corporation\CUDA Samples\v5.0\doc\syntax_highlighting\visual_studio_8\usertype.dat

到

C:\Program Files\Microsoft Visual Studio 11.0\Common7\IDE

目录下

安装VAssistx

下载http://www.wholetomato.com/downloads/default.asp

打开vs2012，依次打开VAssistXX|Visual assist X Options|Projects|C/C++Directories

右侧Platform选择Custom

Show Directorios for选择Other include files

添加

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\include

C:\ProgramData\NVIDIA Corporation\CUDA Samples\v5.0\common\inc

如下图

使Visual Assist X支持.cu文件

打开注册表（win+R，输入 regedit）

找到HKEY_CURRENT_USER\Software\Whole Tomato\Visual Assist X\VANet11

在右侧找到ExtSource，右击修改，添加

.cu;.cuh;

然后重新打开vs

新建一个VC++的Win32控制台项目

附加选项选择“空项目”

• 
  Navigate to the “Configuration Properties\Linker\General” option
• In the “Additional Library Directories” field, add “$(CUDA_PATH)lib\$(PlatformName)”
• Go to the “Configuration Properties\Linker\Input” option
• Lastly in the “Additional Dependencies” field, add “cudart.lib”

  在属性页/链接/常规里找到附加库目录，添加“%CUDA_LIB_PATH%”

  在属性页/链接/输入里找到附加依赖项，添加cudart.lib
  

  
  
  
  

  
  

  最后使用官方给的一个程序测试一下

  #include <iostream>
  
  using namespace std;
  
  __global__ void saxpy(int n, float a, float *x, float *y)
  {
  	int i = blockIdx.x*blockDim.x + threadIdx.x;
  	if (i < n) y[i] = a*x[i] + y[i];
  }
  
  int main(void)
  {
  	int N = 1<<20;
  	float *x, *y, *d_x, *d_y;
  	x = (float*)malloc(N*sizeof(float));
  	y = (float*)malloc(N*sizeof(float));
  
  	cudaMalloc(&d_x, N*sizeof(float));
  	cudaMalloc(&d_y, N*sizeof(float));
  
  	for (int i = 0; i < N; i++) {
  		x[i] = 1.0f;
  		y[i] = 2.0f;
  	}
  
  	cudaMemcpy(d_x, x, N*sizeof(float), cudaMemcpyHostToDevice);
  	cudaMemcpy(d_y, y, N*sizeof(float), cudaMemcpyHostToDevice);
  
  	// Perform SAXPY on 1M elements
  	saxpy<<<(N+255)/256, 256>>>(N, 2.0, d_x, d_y);
  
  	cudaMemcpy(y, d_y, N*sizeof(float), cudaMemcpyDeviceToHost);
  
  	float maxError = 0.0f;
  	for (int i = 0; i < N; i++)
  		maxError = max(maxError, abs(y[i]-4.0f));
  	cout << "Max error: " << maxError;
  }

  然后run吧

  
  

  
  

  参考：http://wenku.baidu.com/view/6ec6b17402768e9951e738d6.html

  http://blog.norture.com/2012/10/gpu-parallel-programming-in-vs2012-with-nvidia-cuda/
  

  
  
• 
  Navigate to the “Configuration Properties\Linker\General” option
• In the “Additional Library Directories” field, add “$(CUDA_PATH)lib\$(PlatformName)”
• Go to the “Configuration Properties\Linker\Input” option
• Lastly in the “Additional Dependencies” field, add “cudart.lib”

  在属性页/链接/常规里找到附加库目录，添加“%CUDA_LIB_PATH%”

  在属性页/链接/输入里找到附加依赖项，添加cudart.lib