RFC: using clz for comparing to zero on the tree level.

[Andrew Pinski]

I accidentally send the email before I was finished.

On Friday, May 9, 2003, at 10:46 US/Eastern, Andrew Pinski wrote:
>
> I think I have a patch for the rs6000 machine description that changes 
> seq to do the expand (I have bootstrapped on ppc-darwin), the only 
> problem is that it produces slower code for (i==0)+a ().
>

(the old way is faster and smaller):
         subfic t,i,0
         addze out,a

(with clz):
         cntlzw t,i
         srwi t1,t,5
         add out,a,t1

This can be fixed by a peephole though (I have not done it yet because 
I am busy with classes lately).

Thanks,
Andrew Pinski

>
> Thanks,
> Andrew Pinski
>
>