Comparison of Fast Inverse Square Root Algorithms for 3D Object Rendering using Ray Tracing

Abstract

The purpose of this paper is to give an insight on the fast inverse square root that is used to compute the inverse square root of a floating-point number to normalize vector that is necessary in many computational problems such as lighting, reflection, and ray tracing. Giving an improvement of speed over 3 times the normal computation way, and an approximation with only 0.1 % error margin. This paper will be comparing several modifications of the algorithm that exist using ray tracing of a spherical object to see which is faster and more accurate through a 3D image rendering and table representation.