Rquest for timings

bearophile bearophileHUGS at lycos.com
Sat Nov 26 02:53:01 PST 2011


Andrea Fontana:

> About "^^": have you tried to benchmark
> 
> 1/(x ^^ n) vs x ^^ -n ?  (with n>0)
> 
> On my setup second version is very very slow.

Take a look at the produced assembly code :-)
Maybe x ^^ n is rewritten with a simple code, while ^^-n calls the pow function.

Bye,
bearophile


More information about the Digitalmars-d-learn mailing list