MathGroup Archive 2005

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: multiprocessor vs cache/memory

  • To: mathgroup at smc.vnet.net
  • Subject: [mg53974] Re: [mg53942] multiprocessor vs cache/memory
  • From: Chris Chiasson <chris.chiasson at gmail.com>
  • Date: Sat, 5 Feb 2005 03:17:21 -0500 (EST)
  • References: <200502040912.EAA01120@smc.vnet.net>
  • Reply-to: Chris Chiasson <chris.chiasson at gmail.com>
  • Sender: owner-wri-mathgroup at wolfram.com

The following post is just an opinion and is slightly off the original
topic. Maybe someone else will share their opinion and enlighten me?

I have come to think of Mathematica more as a research tool than as a
large scale number crunching package. Perhaps cryptography is
different, but it seems like tasks that do not take up much memory or
computer time in commercial solvers for, lets say, finite element
analysis take up way much more memory and time in Mathematica. I know
people may be thinking that I should use Mathematica's sparse solvers
and arrays for memory and speed improvements. I say to them, you are
right! Commercial solvers use those methods by default -- the problems
people typically solve in FEA require those methods to have any chance
of running on a PC. Then we move to problems of interface --
Mathematica has a wonderful front end, but it is not a good CAD/CAE
interface.

None of the above is really a problem. There is much to be said about
using the right tool for the job at hand. Still, I think it is human
to yearn for one tool to do everything.
Regards,

On Fri, 4 Feb 2005 04:12:30 -0500 (EST), joe haydn <nospam at nospam.com> wrote:
> hi folks, we're trying to resolve the classic issue of trade-off between
> multiprocessor vs. cache/memory for a mathematica platform.  Wondered if
> anyone has any kind of input.
> 
> Whether to have more memory/cache (with 3 processors) or to have 4 processor
> machine. (with linux as the OS)
> 
> Or in general thoughts about mathematica use (especially in cryptography
> related computations in our case), I wonder how people evaluate the platform
> on which it runs, performance issues with this tradeoff.
> 
> Any inputs on it, or links to somewhere else for resources to get more
> insights on, I appreciate!
> 
> thx
> joe
> 
> 


-- 
Chris Chiasson
Kettering University
Mechanical Engineering
Graduate Student
1 810 265 3161


  • Prev by Date: NDSolve and differential equation system
  • Next by Date: Few notes
  • Previous by thread: multiprocessor vs cache/memory
  • Next by thread: Re: crash on x86_64: mathematica