Re: [RFC][PATCH] Faster generic_fls

[Daniel Phillips <dphillips@sistina.com>]

On Wednesday 30 April 2003 22:59, Willy Tarreau wrote:
> On Wed, Apr 30, 2003 at 09:15:33PM +0200, Daniel Phillips wrote:
> > In the dawn of time, before God gave us Cache, my version would have been
> > the fastest, because it executes the fewest instructions.  In the misty
> > future, as cache continues to scale and processors sprout more parallel
> > execution units, it will be clearly better once again.
>
> Daniel,
>
> I must acknowledge that your simple code was not easy to beat ! You can try
> this one on your PIII, I could only test it on an athlon mobile and a P4.
> With gcc 2.95.3, it gives me a boost of about 25%, because it seems as gcc
> cannot optimize shifts efficiently. On 3.2.3, however, it's between 0 and
> 5% depending on optimization/CPU.

Was something ifdef'd incorrectly?  Otherwise, there is something the PIII 
hates about that code.  I got 107 seconds on the PIII, vs 53 seconds for my 
posted code at O3, and virtually no difference at O2.  (gcc 3.2.3)

Regards,

Daniel