Path: utzoo!attcan!uunet!jarthur!wilkins
From: wilkins@jarthur.Claremont.EDU (Mark Wilkins)
Newsgroups: comp.sys.mac.system
Subject: Re: MacClassic & System 7.0
Message-ID: <8552@jarthur.Claremont.EDU>
Date: 21 Sep 90 02:03:37 GMT
References: <12280@accuvax.nwu.edu> <70500035@m.cs.uiuc.edu> <1990Sep21.013544.28362@ux1.cso.uiuc.edu>
Organization: Harvey Mudd College, Claremont, CA 91711
Lines: 32

In article <1990Sep21.013544.28362@ux1.cso.uiuc.edu> resnick@cogsci.uiuc.edu (Pete Resnick) writes:
>**FLAME ON**
>It would take a person in a computer science department.....
>The only reason that software becomes unusable on a 4MB machine
>is because developer's have forgotten how to write small code. An
>inordinate number of applications that are out today are just plain
>PIGS. 4MEG???? 32K was a big damn deal 10 years ago; 640K in the
>PC made me drool.

  The reason that Mathematica is so memory-hungry is that by its very nature
it has to perform THOUSANDS of token evaluations for every calculation.  It
is almost entirely symbolic in nature.  As a result, it tends to be slow but
very flexible for the types of math it's good at.

  Mathematica, like most other symbolic manipulation programs, is an extreme
example of speed optimization at the expense of space optimization.  In the
past, time was cheap and memory was expensive.  These days, that is no
longer the case.  Mathematica COULD be jammed into much less space but to
make it faster it generates MASSIVE hash tables of symbols.  Also, all the
graphics are done in PostScript to make them machine independent, at the
expense of space.

  By the way, Maple is more space efficient and somewhat faster, but it also
suffers from the same tradeoff.


  Yes, by yesterday's standards Mathematica is absurdly large.  However, it
would NOT HAVE BEEN POSSIBLE to create such a thing (with all the graphics
capabilities, built-in numerical algorithms, and speed) before multiple-
megabyte systems existed.

-- Mark Wilkins