Optimizing hypot()

Richard Henderson rth@redhat.com
Wed Jul 30 05:12:00 GMT 2003


On Thu, Jul 24, 2003 at 12:27:50AM +0200, Richard Guenther wrote:
>         cvtsi2ss        %eax, %xmm1     # tmp64,
>         movss   %xmm1, -4(%ebp) #,
> 
> Note the bogous temporary xmm1 use at the end.

Note that this is an int->float conversion.

By not including math.h, you lose the prototype for the
function, which means that by C rules the function is
assumed to return int.


r~



More information about the Gcc mailing list