🚀 go-pugleaf

Ian S. Nelson wrote:
> 
> Robert Salasidis <salasidis@medcor.mcgill.ca> writes:
> 
> >Robert Barris wrote:
> 
> >> Mathematica is almost entirely integer based, especially when it comes to
> >> symbolic work and arbitrary precision math as in the factorial example
> >> above.
> >>
> 
> >In the preceeding example using a Sin, I would find it hard to believe that no floating point
> >was used. I know that mathematica does make some use of floating point since I remember their
> >adds that charging an extra 500-1000$ for a version that supported the 80387 coprocessor in
> >the 386 days.
> 
> It was a weighted test... I'm not entirely sure that it had much us either.
> Anyone have a very large spread sheet or mathmatica program?
> 
> >Is Mathematica offered in a 32 bit vs 16 bit version (My reseller catalogs from Ingram Micro
> >and Merisel don't distribute Wolfram Research products so I have no idea) ? If so I would
> >like to know which version was tested.
> 
> He was comparing a Pentium to a 604 right?  I doubt it wuold make much
> difference.  The 604 is for most practical purposes a P6.  It benches a little
> slower and is clocked a little slower but most if not all of the application
> tests I've seen put it a little above the P6.  It will almost always out do a
> Pentium.

It would make a difference if the compiler used was a 16 bit vs 32 bit 
for a program that makes use of > 64k arrays. If the Pentium spends all 
its time doing segment regiter loads then it can become up to an order 
of maginitude slower. I would simply like to know if the version of 
Mathematica is a 32 bit version.

And by the way the 604 is a somewhere between a Pentium and a Pentium 
Pro in speed (excluding the newest Pentiums which at 166MHz do gain some 
advantage). This is assuming that the Pentium Pro is running 32 bit 
apps (ie Spec95 benchmarks etc).