4.9. Floating-Point Division

Floating-point division operations can be costly. Often, a division operation results in a run-time-support call to a predefined function that implements floating-point division. Such calls prevent software pipelining.

If your code divides by a constant that is known at compile time, consider pre-calculating the1/constant value and replacing the division operation with a multiplication by 1/constant. The compiler automatically performs this optimization only if the 1/constant value can be precisely represented in an IEEE-754 float or double.