Xref: utzoo sci.math.symbolic:1624 comp.sys.mac.apps:1147
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!sdd.hp.com!decwrl!shelby!portia.stanford.edu!jessica.stanford.edu!xinwei
From: xinwei@jessica.stanford.edu (Xin Wei Sha)
Newsgroups: sci.math.symbolic,comp.sys.mac.apps
Subject: Re: Mac and Mathematica:  Is an SE/30 enough?
Message-ID: <1990Aug14.215807.21341@portia.Stanford.EDU>
Date: 14 Aug 90 21:58:07 GMT
References: <22758@muvms3.bitnet> <SMITHW.90Aug10121857@hamblin.hamblin.byu.edu>
Sender: news@portia.Stanford.EDU (USENET News System)
Organization: Academic Information Resources
Lines: 18


One good model is to distribute Mathematica across a local network:
Run the Front End on a Mac and the remote kernel on a Sun (or NeXT).
I've tested out various combinations of Macs (II, IIci, cx, fx), with
kernel on Suns (3 & 4 Sparcstation), and NeXTs.  You have the advantage of
an integrated symbolic/graphic notebook interface (which can run in a fixed
memory of 2-3 meg), coupled to a kernel running on a fast machine with
virtual memory.  There ought to be $ savings as well.  We'd like to negotiate
with WRI for a "distributed" license: either many copies of Mathematica
on Macs (perhaps Suns), or multiple Front Ends off of a few Kernels
on some compute servers on the university ethernet.

By the way, on a MacIIfx, compute+graphics timings are comparable to
the NeXT, because, though the NeXT still beats the fx in the
symbol crunching by roughly 9:7, postscript to screen speed is much faster
on the Mac.

Xin Wei