This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patches] Re: i386 multiple-jumps fp comparisons
> After block reordering you have
>
> if (ordered) goto L3; // prob 10%
> L2:
> block_a;
> target:
> block_b;
> ...
> L3:
> if (ge) goto L2; // prob 90%
> goto target;
>
> Surely you can't do any better than that.
I was trying to supply -freorder-blocks to get such sequence, but somehow
it don't work. I will try to figure out why.
(the opcode heruisitics should predict the branches properly, so something
is wrong elsewhere.)
Honza
>
>
>
> r~