[PATCH] improve cexp performance for imaginary inputs

[Joseph Myers]

On Tue, 3 Mar 2015, Julian Taylor wrote:

> Interestingly there is no really significant effect for the same change
> in cexpf and cexpl, I assume this is either due to a faster exp or
> slower sincos. So only double is changed in this patch.

I don't think it's a good idea to get the versions for different types out 
of sync, especially not based on architecture-specific reasoning about the 
speed of particular functions at a particular time on a particular 
configuration.  If anything I'd like to move towards these <complex.h> 
functions being pseudo-templates using macros to have a single main 
implementation shared by all types.

-- 
Joseph S. Myers
joseph@codesourcery.com