Article-I.D.: caip.872
Posted: Thu Jan  2 06:23:37 1986
Date-Received: Fri, 3-Jan-86 01:48:43 EST
Sender: daemon@caip.RUTGERS.EDU
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 165

From: "ROBERTS, JOHN" 


 >  From: enmasse!mroddy@caip.rutgers.edu (Mark Roddy)
 >  
 >>  Several Points of Comparison in the Amiga vs ST Debate
 >>  
 >>  1) The official Atari price of the ST system that has received so much 
 >>  praise in this mailing list is actually $1798 (799 for the monochrome
 >>  version, plus 999 for the color version). 

 >  Hold on there, you can buy a monchrome system (list $799,) and you can buy a 
 >  color monitor seperately for about $400. Total price: ~$1200. That is the
 >  same price as an Amiga without any monitor at all. The monochrome version

I assume you mean that Atari/authorized dealers have decided to sell the
color monitor as a separate item. If this is correct, I am glad that 
Atari has decided to be sensible about the matter. I had read in several
publications that it was not possible to purchase a monitor as a 
separate item, and that was one of my main complaints about the 
marketing of the ST. I understand that normal NTSC-compatible RGB 
monitors cannot be used with the ST. Please correct me if I'm wrong.


 >>  4) Commodore/Amiga has been fairly open and honest about the internal
 >>  structure and functioning of their machine, while by comparison 
 >>  Atari has been relatively secretive (specific examples furnished upon 

 >  Uh, I have the schematics at home, they came with the developers kit,
 >  availabler to the general public for $300.

Could you please post a name and address by which such information might
be purchased? (Admittedly shaky justification for posting to an Amiga
mailing list, but since a number of Amiga users are also interested
in the ST, I think it might be forgiven just this once.) In a product
survey conducted by reading every Atari-oriented, Commodore-oriented,
and general interest computer magazine I could find, I was able to
find out a lot more about the internal structure and functioning of
the Amiga than of the ST. Maybe the degree of willingness to communicate
with publishers/reporters has been the main reason for the difference
in coverage, rather than desire for secrecy.


 >>  6) Most of the Dhrystone benchmarks run on the Amiga (showing slightly
 >>  slower execution than the ST) have been run with the processor using
 >>  the same memory as the specialized chips. With an expanded memory
 >>  space (which could be desired for its own sake) performance should be
 >>  enhanced, especially for graphics-intensive applications. A faster
 >>  processor (the 68020) should give a greater increase in performance.
 >>  What should really give floating point performance a boost is the
 >>  addition of a good floating point processor chip (of which the 68881
 >>  is a logical choice for the 68000 family). I expect a 68881 to be
 >>  available for the Amiga long before the ST gets one (if ever).
 >  
 >  Wait, hold on here. Memory outside of the lower 256K on the Amiga is
 >  slower, not faster. The Dhrystone benchmarks are CPU only, not 
 >  graphics intensive.

 >  Hell, if we redesign the ST it will be better also.

Incorrect. The basic machine comes with 256K RAM, and another 256K
can be added for about $200. (There is a separate non-user-accessible
RAM to hold the operating system in current machines, unlike the
ST, which eats up a big chunk of user RAM to hold the OS in current
machines.) The Amiga treats the "internal" 512K user address space
as a contiguous and uniform chunk of memory which is shared by the
processor and the special-purpose graphics, sound, and I/O chips.
This memory is designed to be faster than needed for any particular
device in order to minimize contention (it is effectively a 
time-slice multiport memory), but for some applications there will
still be a certain amount of contention, thus slowing down the
processor. The Amiga is designed to make it easy to add external
memory, which is accessed only by the processor. Processes running
in external memory (still in the same linear address space, of
course) are not subject to slowdown from memory contention, and
will therefore be as fast, or sometimes significantly faster than
processes in the "chip memory".

I mentioned graphics in reference to Dhrystones because there are
certain applications that are both computation- and graphics-
intensive, and because the computational performance of the Amiga
(and, I suspect, the ST) can be affected by what is being displayed.

I am thinking of the 68881 not as a redesign, but as an add-on
component. The 68881 is a very expensive ($240 - EDN 11/28/85)
and, until very recently, nonexistent chip, but its eventual release
was announced quite a while back. While it would not have been 
logical to design either the Amiga or the ST with a 68881, it would
have been sensible to design the machines so one could be added later.
With its comprehensive external port, I think Amiga has done this.

I assume that both companies are thinking of eventually making improved
versions of their machines, with faster processors, etc. The
piggyback board with 68020 and 68881 is an unexpectedly early upgrade.


 >>  7) A minimal Amiga system I would buy (512K, RGB monitor) costs
 >>  considerably more than a minimal Atari 520ST system I would buy.
 >>  I think the Amiga would better suit my intended applications.

 >  Even with your suggestion to buy two ST's, the Amiga costs more.

I think one Amiga would better suit my intended applications than
two ST's. An Amiga AND an ST would be even nicer.


 >>  9) The fact that the markets for the two machines do not completely
 >>  overlap may help both companies to survive. My perception of local
 >>  market conditions is that STs, Amigas, and C-128s (surprisingly) are
 >>  being sold as fast as the dealers can get them, at least for the
 >>  time being. Perhaps market share will be determined by which company
 >>  can produce computers most quickly.

 >  I haven't done a market survey, but at the local high volume retailer,
 >  (Bit Bucket, Newton MA,) the Amigas are stacked up to the ceiling, and
 >  they don't have an inventory of STs. (They report that the STs sell as
 >  fast as they come in, the Amiga business is slow and steady.) They
 >  also say that lack of Amiga software  while the ST software is quite
 >  impressive, has had a severe impact on Amiga sales.

The first printed information on Amiga sales I have been able to find
is a note in the December 30 issue of Electronic News concerning unit
market share of personal computers for November, from data collected
by InfoCorp. It lists the Amiga as having 4% of the November unit market
share, compared to 7% for the Macintosh, 8% for the IBM PC/AT, 13% for
the Apple IIc, etc. Not bad for a machine barely out of the starting 
gate, comparatively speaking. The ST is not specifically listed, and
since EN has covered the ST in previous issues, I do not have any
reason to believe that it is being ignored. It may fall into the 
category of "All Others (2% or less)" that collectively made up 31%
of the unit market share. Note that these figures do not directly
reveal total revenues or profits, but I suspect that Commodore/Amiga
makes more money per unit sold than Atari.

In reference to total company revenues, I just read an editorial in a
Commodore-oriented (but not -owned) magazine expressing surprise that
the C-128 is selling much better than expected. Commodore also has
a different marketing program (with different machines) in Europe,
which it claims was not hit nearly as hard as domestic marketing during
the recent industry-wide slump.

Still other sources indicate that Atari is making a nice profit selling
the ST, presumably enough to keep the company going.

Does anybody have any statistics on actual numbers of units sold?
-- 
 > 					Mark Roddy
 > 					Net working,
 > 					Just reading the news.
 > 				(harvard!talcott!panda!enmasse!comm!mark)


I think a rational discussion of the issues, such as this appears to
be, will produce a lot more useful information on the two machines
than some of the flames that have been appearing lately.

Sorry about the length of this posting. If you think it's too long, you
don't have to read the whole thing.

(Same disclaimers as last time.)

                                       John Roberts
                                       roberts@nbs-vms.ARPA

------

Article-I.D.: cwruecmp.1365
Posted: Thu Jan  2 21:05:18 1986
Date-Received: Fri, 3-Jan-86 05:32:03 EST
References: <872@caip.RUTGERS.EDU>
Organization: CWRU Dept. Computer Eng., Cleveland, OH
Lines: 116

 >  From: "ROBERTS, JOHN" 
 >  I assume you mean that Atari/authorized dealers have decided to sell the
 >  color monitor as a separate item. If this is correct, I am glad that 
 >  Atari has decided to be sensible about the matter. I had read in several
 >  publications that it was not possible to purchase a monitor as a 
 >  separate item, and that was one of my main complaints about the 
 >  marketing of the ST. I understand that normal NTSC-compatible RGB 
 >  monitors cannot be used with the ST. Please correct me if I'm wrong.

Yes the color monitor is being sold separately (and has been for the
at least the last three months). The ST can be hooked
to an analog RGB monitor, but not to a RGBI (ibm PC flavour
montitors), quite easily (cables and 3 resistors).

 >  
 >>>  4) Commodore/Amiga has been fairly open and honest about the internal
 >>>  structure and functioning of their machine, while by comparison 
 >>>  Atari has been relatively secretive (specific examples furnished upon 
 >  
 >>  Uh, I have the schematics at home, they came with the developers kit,
 >>  availabler to the general public for $300.
 >  
 >  Could you please post a name and address by which such information might
 >  be purchased? (Admittedly shaky justification for posting to an Amiga
 >  mailing list, but since a number of Amiga users are also interested
 >  in the ST, I think it might be forgiven just this once.) In a product
 >  survey conducted by reading every Atari-oriented, Commodore-oriented,
 >  and general interest computer magazine I could find, I was able to
 >  find out a lot more about the internal structure and functioning of
 >  the Amiga than of the ST. Maybe the degree of willingness to communicate
 >  with publishers/reporters has been the main reason for the difference
 >  in coverage, rather than desire for secrecy.
The $300 dev. system can be ordered from Richard Frick, Atari
Corp,1196 Borregas Ave , Sunnyvale CA 94086. Included in the package
is the C compiler, linker, assembler, Uemacs, Libraries, a bunch of
utilities, command line processor, schematics, data sheets for some of
the commercial chips, icon and shape editor, resource construction set
and 2800 pages of docs. So, as you can see the price is in line with
the price of the machine. There is also a book due out soon about the
internals if you don't want the whole dev. system. (Don't ask me when exactly).

 >  
 >  
 >>>  6) Most of the Dhrystone benchmarks run on the Amiga (showing slightly
 >>>  slower execution than the ST) have been run with the processor using
 >>>  the same memory as the specialized chips. With an expanded memory
 >>>  space (which could be desired for its own sake) performance should be
 >>>  enhanced, especially for graphics-intensive applications. A faster
 >>>  processor (the 68020) should give a greater increase in performance.
 >>>  What should really give floating point performance a boost is the
 >>>  addition of a good floating point processor chip (of which the 68881
 >>>  is a logical choice for the 68000 family). I expect a 68881 to be
 >>>  available for the Amiga long before the ST gets one (if ever).
 >>  
 >>  Wait, hold on here. Memory outside of the lower 256K on the Amiga is
 >>  slower, not faster. The Dhrystone benchmarks are CPU only, not 
 >>  graphics intensive.
 >  
 >>  Hell, if we redesign the ST it will be better also.
 >  
 >  Incorrect. The basic machine comes with 256K RAM, and another 256K
 >  can be added for about $200. (There is a separate non-user-accessible
 >  RAM to hold the operating system in current machines, unlike the
 >  ST, which eats up a big chunk of user RAM to hold the OS in current
 >  machines.) The Amiga treats the "internal" 512K user address space
 >  as a contiguous and uniform chunk of memory which is shared by the
 >  processor and the special-purpose graphics, sound, and I/O chips.
 >  This memory is designed to be faster than needed for any particular
 >  device in order to minimize contention (it is effectively a 
 >  time-slice multiport memory), but for some applications there will
 >  still be a certain amount of contention, thus slowing down the
 >  processor. The Amiga is designed to make it easy to add external
 >  memory, which is accessed only by the processor. Processes running
 >  in external memory (still in the same linear address space, of
 >  course) are not subject to slowdown from memory contention, and
 >  will therefore be as fast, or sometimes significantly faster than
 >  processes in the "chip memory".
 >  
In my opinion, allowing the graphics processor to only access the
'internal' memory is rather limiting tradeoff, if, for instance you
are writing applications that want to define frames in the process
address space and want to change the screen base pointer to the frame
(a popular 'trick' ), and so on. $200 sounds a little steep for 256K
of memory, considering that 2Meg addon cards are going for ~$350 for
the ST. But i guess the pricing for peripherals and software for the
Amiga should be expected to be propotionally higher. BTW the OS in
Eprom (this month in rom) is out for the Atari, so now you have lots
of memory.

 >  I mentioned graphics in reference to Dhrystones because there are
 >  certain applications that are both computation- and graphics-
 >  intensive, and because the computational performance of the Amiga
 >  (and, I suspect, the ST) can be affected by what is being displayed.
 >  
 >  I am thinking of the 68881 not as a redesign, but as an add-on
 >  component. The 68881 is a very expensive ($240 - EDN 11/28/85)
 >  and, until very recently, nonexistent chip, but its eventual release
 >  was announced quite a while back. While it would not have been 
 >  logical to design either the Amiga or the ST with a 68881, it would
 >  have been sensible to design the machines so one could be added later.
 >  With its comprehensive external port, I think Amiga has done this.
 >  
 >  I assume that both companies are thinking of eventually making improved
 >  versions of their machines, with faster processors, etc. The
 >  piggyback board with 68020 and 68881 is an unexpectedly early upgrade.
 >  
The current hot rumor is that Atari will be showing its 32 bit
processor based add on box, nickednamed 'TT', at CES. The ST will
supposedly be a front end for this cruncher. I hope they use 68020 and
not a 320XX.
-- 
					Jwahar R. Bammi
			       Usenet:  .....!decvax!cwruecmp!bammi
			        CSnet:  bammi@case
				 Arpa:  bammi%case@csnet-relay
			   CompuServe:  71515,155

Article-I.D.: ukma.2446
Posted: Fri Jan  3 05:05:26 1986
Date-Received: Sat, 4-Jan-86 04:39:25 EST
References: <872@caip.RUTGERS.EDU>
Reply-To: knnngt@ukma.UUCP (Alan Kennington)
Organization: Univ. of KY Mathematical Sciences
Lines: 21


	On the subject of the relative secretiveness of the two machines,
one of the greatest pleasures of the ST is the lack of it, and one of the
most often heard complaints of Amiga owners is the excess of it. So you
have really picked the wrong aspect to grandstand on here.
	I may be wrong, but I have the distinct impression that the lack
of MMU gives the user access to the entire memory space. The space from
$0 to $7FFFF is readable and disassemblable and writeable by a BASIC
program in the ST. Also readable are the deadstart ROM chips at $FC0000
and, I think, $FA0000. I haven't got my developer's kit yet. After only
4 weeks over the Christmas period I am willing to forgive this.
	But if you want a copy of the development system (complete with
so much that I won't bore the good patient readers with the oft-repeated
list), just send $300 + $21(pack/hand/ship) to Atari Corporation,
1196 Borregas Avenue, Sunnyvale, California 94086 Atten. Richard Frick,
with your ST serial number (!). Oh, and by the way, you won't need
character references, and they won't check your criminal record (or will
they? They'd better not!).
	One final point: excessive quotation of recently posted items
is tedious, and so is biased "discussion". The best kind of discussion is
that which elicits NEW (!) FACTS(!!).  (end of speech.)  (ak).