Chapter 4. 2D Grids and Interactive Graphics

In this chapter, we see that the CUDA model of parallelism extends readily to two dimensions (2D). We go through the basics of launching a 2D computational grid and create a skeleton kernel you can use to compute a 2D grid of values for functions of interest to you. We then specialize the kernel to create dist_2d, an app that computes the distance from a reference point in the plane to each member of a uniform 2D grid of points. By identifying the grid of points with pixels in an image, we compute data for an image whose shading is based on distance values.

Once we are generating image data, it is only natural to take advantage of CUDA’s graphics interoperability (or graphics interop for short) capability, ...

Get CUDA for Engineers: An Introduction to High-Performance Parallel Computing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.