> -----Original Message-----
> From: Roman Pearce [mailto:rpearcea at gmail.com]
> Sent: Monday, March 31, 2008 8:43 PM
> To: rjf
> Subject: Re: benchmarking CAS
>
> > Are you saying that if the polynomials were small, Maple
> would not use
> > the same routines and would not run with garbage collection?
>
> That's correct. For small polynomials Maple expands and divides
> products in the kernel, using compiled C code that runs without
> garbage collection. For large problems the kernel calls back into the
> Maple library, to the routines `expand/bigprod` and `expand/bigdiv`
> and `expand/bigpow`. These routines are recursive, interpreted Maple
> code that run with garbage collection. That's why in your paper on
> sparse multiplication Maple 7 was found to be much slower than
> MapleVR4, etc. They had to change it because for their algorithms,
> doing a big problem without garbage collection will crash the system.
>
> On 3/30/08, rjf <fateman at gmail.com> wrote:
> > On Mar 29, 6:28 pm, Roman Pearce <rpear... at gmail.com> wrote:
> >
> > >
> > > Because the polynomials are large, Maple uses interpreted routines
> > > that run with garbage collection.
> >
> > Are you saying that if the polynomials were small, Maple
> would not use
> > the same routines and would not run with garbage collection?
> >
> >
>