Seven Concurrency Models in Seven Weeks

Day 1: GPGPU Programming

Today we’ll see how to create a simple GPGPU program that multiplies two arrays in parallel, and then we’ll benchmark it to see just how much faster the GPU is than the CPU. First, though, we’ll spend a little time examining what makes a GPU so powerful when it comes to number-crunching.

Graphics Processing and Data Parallelism

Computer graphics is all about manipulating data—huge amounts of data. And doing it quickly. A scene in a 3D game is constructed from a myriad of tiny triangles, each of which needs to have its position on the screen calculated in perspective relative to the viewpoint, clipped, lit, and textured twenty-five or more times a second.

The great thing about this is that although the amount of data ...