Hands-On GPU Programming with Python and CUDA

上QQ阅读APP看书，第一时间看更新

Setting up Visual Studio (Windows)

At the time of writing, only one version of Visual Studio appears to ingrate perfectly with both Python and the latest CUDA Toolkits—Visual Studio 2015; that is, Visual Studio version 14.0.

While it may be possible to make a sub-installation of this under a later version of Visual Studio (for example, 2017), we would suggest to the reader that you directly install Visual Studio 2015 with C/C++ support onto your system.

Visual Studio Community 2015, the free version of this software, can be downloaded at https://visualstudio.microsoft.com/vs/older-downloads/.

Here, we will do a minimalist installation, with only the necessary components for CUDA. We run the installation software, and select the Custom installation:

Click Next, then click the drop-down box for Programming Languages, and then choose Visual C++ (feel free to select other packages or programming languages if you want or need them for other purposes, but Visual C++ is all we will need for GPU programming):

This should take some time to install. After this is complete, we will be ready to install the CUDA Toolkit.