5.7.10.6 Relaxed-math Option

The -relaxed-math option links in a relaxed-compliance math library.

This library provides alternative floating-point support routines that are faster and smaller than the default math routines, but make some sacrifices in compliance. For instance, it does not do all of the infinity, overflow and NaN checking, etc., of a fully-compliant library. It does not return all of the detailed feedback from that checking. However, this reduced compliance is usually sufficient for most applications.