[EMAIL PROTECTED] wrote:
> 
> On Wed, 20 Mar 2002, Gordon Pearce wrote:
> 
> > > >
> > > > Why will it be faster?  Is it something to do with the way the array is
> > > > referenced?
> > >
> >
> > From the previous two email, yep, that's pretty much what I thought.
> > I never thought about the cache thing though.  I wonder if it makes a
> > difference with no cache?
> > I think an experiment is in order...
> 
> Well, you could also get a difference from another effect, if your
> arrays are large. Think virtual memory. If your arrays take a number
> of pages larger than the number of slots in the TLB (the TLB is a fast
> cache for page table lookups, ie virtual->physical address lookups),
> then wandering back and forth through memory will cause extra memory
> accesses to re-fetch the entries that are pushed out of the TLB.
> 

Atlas Blas optimises itself on crazy stuff like this.

http://www.netlib.org/atlas/

Quite frankly, it scares me :).

Thanks,

Allan
-- 
Claiming that your operating system is the best in the world because
more
people use it is like saying McDonalds makes the best food in the world.
--------------------------------------------------------------------
http://www.lug.org.uk                   http://www.linuxportal.co.uk
http://www.linuxjob.co.uk               http://www.linuxshop.co.uk
--------------------------------------------------------------------

Reply via email to