MathGroup Archive 2000

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

Search the Archive

Re: Help! Mathematica on my 500MHz outperforms my GHz machine!

  • To: mathgroup at smc.vnet.net
  • Subject: [mg21844] Re: Help! Mathematica on my 500MHz outperforms my GHz machine!
  • From: paulh at wolfram.com (P.J. Hinton)
  • Date: Sun, 30 Jan 2000 18:53:27 -0500 (EST)
  • Organization: "Wolfram Research, Inc."
  • References: <86158s$jss@smc.vnet.net> <870ujq$8kb@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

In article <870ujq$8kb at smc.vnet.net>,
 terryis at my-deja.com writes:

> Of the more notable responses, RogerWilson (roger.wilson at nomura.co.uk)
> wrote to say that Mathematica 3 does all floating point calculations
> in software and that the supposedly very fast floating point unit in
> my Athlon was therefore not being used (as far as he knew).
> 
> Is this true? I mean no disrespect to Roger but I had always thought
> that Mathematica uses both arbitrary and machine precision depending on
> the characteristics of the number in question.

As long as your inputs specify machine precision numbers, Mathematica
typically uses machine precision arithmetic to achieve its results.  
It will make use of the floating point capabilities of your CPU.  This
is noted in Section 3.1.6 of _The Mathematica Book_ Fourth Edition.

http://documents.wolfram.com/v4/MainBook/3_01/S3.1.6.html

Note, however, that this does not include CPU-class-specific 
optimizations such as Intel's MMX instructions or AMD's 3DNow!
instructions.

http://support.wolfram.com/Systems/Windows/Compatibility/MMX.html

-- 
P.J. Hinton
Mathematica Programming Group           paulh at wolfram.com
Wolfram Research, Inc.
Disclaimer: Opinions expressed herein are those of the author alone.


  • Prev by Date: Re: warning: action not found: delete-next-character
  • Next by Date: Re: CORRUPTED NOTEBOOKS
  • Previous by thread: RE: Help! Mathematica on my 500MHz outperforms my GHz machine!
  • Next by thread: Flat, OneIdentity Again