[Gllug] Re: [OxLUG] C function strcasecmp

[Alex Hudson]

On Sun, Jul 22, 2001 at 08:43:34PM +0100, Nix wrote:
> *selfLART*

Careful now :)

> >                                    Shifts should be _much_ faster than
> > multiplies on most Intel chips,
> 
> I'm not sure *what* chip I was thinking of. I'll have to check the GCC
> sources :)

Certainly, gcc won't optimize multiplies to adds-and-shifts on the P4
because they're slower - Intel needed to do some transistor chopping, IIRC,
and implemented the barrel shifts in the ALU, hence each shift is worth a
muiltply. 

Leads to interesting comparisons - most processors complete a shift in one
clock tick, the P4 (notwithstanding its pipelined-ness) takes 19 :)

Cheers,

Alex.

-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug