GPGPU Programming for Games and Science

Chapter 2

CPU Computing

2.1 Numerical Computing

The projects I have worked on have a common theme: numerical computing

using floating-point arithmetic. It is quite easy to compute numerically using

the hardware-supported

float and double data types. The process is simple.

Study the mathematics for the problem at hand, develop an abstract algorithm

for the solution, code up a few equations, and then compile, link, and execute—

the results are at your finger tips. Does the algorithm require computing the

roots to a quadratic equation? No problem, just use the quadratic formula.

Do you need to compute the distance between two line segments? Again, no

problem. ...