Operating System (OS): Windows 7 set into the development environment (IDE): Microsoft Visual Studio 2008 SP1 CUDA version (CUDA version): 3.0
Hardware that supports CUDA when CUDA programming is not necessary, and Cuda provides a way to simulate GPU operations with CPUs, so Cuda programs can be written on any configured machine. However, to experience cuda large-scale parallel high-speed operations, it is necessary to support CUDA hardware.
DownloadCuda Downloads: http://developer.nvidia.com/object/cuda_3_0_downloads.html cuda VS Wizard Downloads: http://sourceforge.net/ projects/cudavswizard/
Installation StepsThe general procedure for installing Cuda is to install the video card driver, install Toolkit, that is, Cuda Toolkit, install SDK,GPU Computing SDK, configure environment variables and IDE variables. Detailed steps
1. Install the graphics driver, you can find the latest drive on the Cuda download page.
2. Install Toolkit SN Custom to install it in C:/cuda/toolkit location.
3. Install the SDK (GPU Computing SDK) SN Custom to install it in the C:/cuda/nvidia GPU Computing SDK location, so that it is easy to find the files inside.
4. Syntax highlighting 4.1 copies the%nvidia GPU Computing sdk%/c/doc/syntax_highlighting/visual_studio_8 under NVIDIA's written grammar file Usertype.dat to% Visua studio%/common7/ide 4.2 Start Visual Studio, select Tools > Options > Text editor > File extensions, set to CU with the editor select Microsoft Visual C + +, click "OK" and 4.3 reboot visual Studio.
5.IDE environment variable setting starts visual Studio, selects Tools > Options > Projects and Solutions > "C + + catalog", platform selects your operating system's corresponding platform, and adds a path to include files%toolkit%/ Include and%nvidia GPU Computing sdk%/c/common/inc add Paths%toolkit%/lib,%toolkit%/lib64, and%nvidia GPU Computing in library files sdk%/c /common/lib add paths in source files%toolkit%/src and%nvidia GPU Computing sdk%/c/common/src 6. Add 6.1 of compilation rules use CUDA vs Wizard CUDA vs Wizard is a Openhero-open, Open-source, Visual Studio-oriented tool. When the installation is complete, a Cudawinapp template appears when you create a new project in Visual Studio, which eliminates many settings; 6.2 Manually Configure compilation rules after creating a new Win32 console application, the Project menu > Custom Build Rule > "Find existing" > find compiled rules found in Cuda.rules > "OK" > Check%nvidia GPU Computing Sdk%/c/common. Add "Cudart.lib cutil32D.lib" in the Project menu > Properties > Linker > Enter > "Additional Dependencies".
Once the configuration is complete, you can start writing the CUDA program.
--
Snigoal,hust,wuhan,china
Reprint Http://snigoalx.spaces.live.com/blog/cns!5DA334179A108A0A!228.entry
| First, there are many similar methods on the Internet, some complex, some simple, some not necessarily can be achieved, the method here in the 64-bit Win7 system has been installed to achieve the operation, which there are some possible problems for everyone to solve. You can refer to this colleague to write the configuration process, basically can succeed: Http://snigoalx.spaces.live.com/blog/cns!5DA334179A108A0A!228.entry. And I won't go into it. But generally we will find that if the 64-bit system, open VS2008 Run the program, our VS2008 (i installed this version) but can not go to debug, that is, or in the solution platform solution without x64, or even with x64, But when you choose, you find that you can't debug. This is because we did not install the VS2008 64-bit compiler, this situation we have a few people have encountered, may be because there was no installation or VS2008 itself did not install the default, this time we need to upgrade, install 64-bit compilers themselves. But we don't need to go back to the 64-bit compiler to download the installation again, just turn on the computer----uninstall or change the program, find VS2008 installation program, click Uninstall, in fact, this time is not uninstall, but will remind you is upgrade or uninstall, we upgraded VS2008 on the line, In the upgraded directory, we will find that there are 64-bit compilers installed, and it will be installed. Another possible scenario for is "Unable to start the program because the computer lost Cutil64d.dll", we just need to find cutil64d.dll and then copy it to the System32 folder under C disk. In this way, in the 64-bit Win7 environment Cuda installation configuration is basically completed, you can implement our Cuda program. |
