Re: FastMathTestPerformance results misleading

2011-03-31 Thread sebb
On 1 April 2011 01:23, Bill Rossi wrote: > > This test shows that Math.sin() is faster than FastMath.sin(), but this is > misleading because the test operates over the domain 0 < x <1, whereas legal > arguments to sin() are all finite numbers.   In particular, its when |x| is > large that performa

FastMathTestPerformance results misleading

2011-03-31 Thread Bill Rossi
This test shows that Math.sin() is faster than FastMath.sin(), but this is misleading because the test operates over the domain 0 < x <1, whereas legal arguments to sin() are all finite numbers. In particular, its when |x| is large that performance of the Math trig functions really drops off