Path: utzoo!utgpu!watmath!att!mtuxo!mtgzz!drutx!druwy!dlm
From: dlm@druwy.ATT.COM (Dan Moore)
Newsgroups: comp.sys.atari.st
Subject: Re: CMI Accelerator Board
Message-ID: <4202@druwy.ATT.COM>
Date: 24 Jul 89 15:54:04 GMT
References: <812@uhnix2.uh.edu>
Organization: AT&T, Denver, CO
Lines: 55

in article <812@uhnix2.uh.edu>, uace0@uhnix2.uh.edu (Michael B. Vederman) says:
> Well, after having waited with bated breath for the CMI accelerator board, I
> would like to share my experiences with the rest of the net...
> 
> *** FLAME ON ***
> [ long description of CMI's shipping problems (they didn't ship when
>   they said they did) and the failure of the accelerator in 2 different
>   STs. ]
> *** FLAME OFF ***

	I'm not surprised that CMI's accelerator didn't work in your
STs.  I've seen several beta copies of accelerators for the ST, all of
them are flakey in at least some STs.  Some are flakey in all STs.  I
have seen the CMI accelerator work in an ST, I've also seen it fail.

	This is due to the design of the ST, Atari saved you (the
buyer) money by building machines that just barely run.  It takes very
little to push an ST out of spec so that it will fail.  To make things
even more interesting there are "many" different STs, different revs of
motherboards, different brands of chips (with different bugs in some of
the chips).  This means there really isn't an "average" ST that you can
design hardware for.  And even if the board works in the ST you may not
be able to talk to the outside world (hard disk/floppy disk) since most
ST device drivers use timing loops.  Fortunately many of the loops are
several times longer than they need to be, so it doesn't matter if they
run faster.

	When I was at dP (back in the bad old days) we tested the
Translator One with a lot of different STs and drives (about 15 machines
between the office and the beta-testers).  We thought all the problems
were solved but when we shipped we found that it wouldn't work with a
large fraction of the STs in the world.  I suspect that CMI (and the
other builders ST accelerators) have the same problem, it worked in all
the STs they tested with but that has very little to do with working in
the real world.

	Hopefully CMI, and the other accelerator companies, will be
willing to give purchasers a 100% refund if the board fails.  There is
one ST hardware company that refuses to give a refund if their disk
enhancement fails to work in the purchasers ST, hopefully the
accelerator companies won't adopt the same attitude.

	Personally I would spend a LOT of time thinking before deciding
to buy an accelerator for the ST.  There is a good chance that it may
not work in any given ST.  Also you may not gain very much, some
accelerators only give you a 5 to 10% boost, others can give a 50 to
100% boost.




				Dan Moore
				AT&T Bell Labs
				Denver
				dlm@druwy.ATT.COM