Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!convex!bcm!lib!thesis1.med.uth.tmc.edu
From: jsorenso@thesis1.med.uth.tmc.edu (JEFFREY MARIUS SORENSON)
Newsgroups: comp.sys.mac.hardware
Subject: Fastest Mac FP Setup?
Message-ID: <5146@lib.tmc.edu>
Date: 24 Jun 91 23:51:43 GMT
Sender: usenet@lib.tmc.edu
Distribution: usa
Organization: University of Texas Medical School at Houston
Lines: 33
Nntp-Posting-Host: thesis1.med.uth.tmc.edu
Originator: jsorenso@thesis1.med.uth.tmc.edu

The lab I'm working for wants to get a Mac to run some floating-point
calculation intensive simulations (with a friendly interface of course).
What is the fastest combination available?  I'm guessing that a ][fx with a
Radius Rocket is the fastest dynamic duo.  Is there a faster combination?
Other concerns:
1. I read a posting here about the '040 that said FP operations on the '040
require a 256K block of code to translate floating point instructions
for the '030 to the corresponding '040 instructions.  Is this the case?  If
so, is this necessarily a run-time translation or are there utilities to let
you recompile your code so that the translation need only be done once.
Along these lines, are there ANY compiliers/utilities for that let you
exploit the floating point power of the '040.  Does the Rocket come with any
such tools?  If not, and it is the case that run-time translation must
occur, then is the Rocket REALLY 3x faster than a stock ][fx (or only
theoretically 3x faster) at floating-point stuff?  Is there anyone out
there who's played with both a ][fx and a Rocket who could give advice?

2. If there's a Rocket under the hood, is there any advantage (as far as
numerical simulations go) to having an fx as opposed to a ci? (wouldn't
both CPUs be converted into I/O processors?).  Will a Rocket even fit
into a ci?

3. Not a major concern, but it would be nice if the machine were sys 7
compatible.  I've heard that the rocket is almost there, but will it work
at all?  Is it just a problem virtual memory and addressing or are there
a lot of patches involved?

Thanks in advance for any insight.
-- 
-------------------------------------------------------------------------------
jsorenso@thesis1.med.uth.tmc.edu | "No matter where you go, there you are"
				 |				-BB
-------------------------------------------------------------------------------