Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site cvl.UUCP
Path: utzoo!linus!decvax!decuac!cvl!gyuri
From: gyuri@cvl.UUCP (Gyorgy Fekete)
Newsgroups: net.micro.atari
Subject: In defense of J. Demar (Bit longer than I wanted)
Message-ID: <869@cvl.UUCP>
Date: Fri, 11-Oct-85 13:30:51 EDT
Article-I.D.: cvl.869
Posted: Fri Oct 11 13:30:51 1985
Date-Received: Sat, 12-Oct-85 08:24:46 EDT
Distribution: net
Organization: Computer Vision Lab, U. of Maryland, College Park
Lines: 80

In response to the flame to John Demar's comments,

John Demar (JD):
	The ST brings in a 32K file
	in less than 4 seconds, including drive start up,
	directory search, etc. The Amiga takes almost 20 seconds!!

Gary Girzon (GG):
	That is just not true! The BYTE Basic benchmark for reading and
	writting a 64k file takes 25 seconds. And this is in basic!
	Please check or tell us how you arrived at these figures.

JD:	I'm a software/hardware developer and an electronics engineer.
	I've seen and used computers from $50 to $5Million and have designed
	VLSI chips for 6 years at GE until starting my business last year.
	So, the following is said from technical expertise and not
	first-impression judgments from marketing "fluff"...
	I own (or have owned) both the Atari 520ST and Amiga PC. I've given
	them both a good bit of work and inspection, including o/s design and
	hardware architecture. Here are some facts and my conclusion:...

I:	Under these circumstances I tend to believe Demar more so than BYTE.
	Without citing, I have read some of Byte's previous benchmarks, that
	were misleading, but this is really irrelevant.
	By the way, Is this J. Demar the same J. Demar who wrote some of those
	Blackburg series books?

JD(cont):...the amiga has a non-standard disk configuration and does
	much of the disk support in software (ie. slow). The drives have
	slightly more capacity than the the ST's DS drives (880K to 720K) but
	this is at the expense of speed. The Amiga directory format (or
	lack of) is done much like a commodore-64. In fact, to get a
	directory, the Amiga goes out and finds a program called DIR,
	loads it and goes back searching!

I:	If that is true, I have lost all respect for AmigaDOS.

GG:	...Unless the ST has some sort of dual-ported RAM, the screen data
	must come from the same data bus in the ST as well.

JD:	..First, there are a pair of chips working together to optimize
	data bus and screen data access. The memory controller
	prefetches 16-bit data directly for the 68000 and also places
	screen data onto a separate bus for the screen refresh chip. This
	operation only steals 8-18% of the available true CPU time.

GG:	If you measure a computer's worth by the number of chips,
	you might as well pick an IBM over a Mac. Perhaps the custom chips
	in the AMIGA are  better integrated to perform several functions in one
	chip. Thus you do not need MORE chips if three can do the job. 

I:	non sequatur (sp?)

JD:	In the low res mode, those fantastic graphics chips
	steal almost 70% (yes!) of the possible CPU time that the 68000
	could be using to do real computer things like calculate,
	move/sort data...

GG:	How does one arrive at the magic 70% time figure? Also, the 68000
	in the Amiga does not draw lines, polygons or fill areas.
	It is done by one of the coprocessors. SO the 6800 (???) can
	do "real" computer things.

I:	I would also like to know how Demar got the 70% figure.

JD:	The windows are poorly configured and move with flicker.

GG:	They do? Not in my AMIGA. 

I:	Maybe they do in his AMIGA.

Sorry for the redundancy in the quotes.
-- 
Gyorgy Fekete --- University of MD, Computer Vision Lab, (301) 454-4526

gyuri@cvl.{ARPA,CSNet}

...seismo  \
...allegra +--  !umcp-cs!cvl!gyuri.UUCP
...brl-bmd /