NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/sys/amiga/graphics/5983

Internet Message Format | 1992-09-13 | 8.9 KB

Xref: sparky comp.sys.amiga.graphics:5983 comp.sys.amiga.hardware:15250
Newsgroups: comp.sys.amiga.graphics,comp.sys.amiga.hardware
Path: sparky!uunet!utcsri!torn!spartan.ac.BrockU.CA!aciceran
From: [e-mail] (A. J. Ciceran)
Subject: CBM Discusses Future Amigas
Message-ID: <[e-mail]>
Organization: Brock University, St. Catharines Ontario
Date: Sun, 13 Sep 1992 16:18:33 GMT
Lines: 188

[Since much of the information in the following post is of particular
 interest to those involved in graphics/multimedia, I have cross-posted
 this message from c.s.a.misc. If you've already ventured into c.s.a.misc
 or c.s.a.advocacy, you've probably already encountered it. Special
 thanks goes out to Andrew Yew for providing a thorough overview of the
 WOC events and information sessions.]

--
    Wow!

    I just got back from the second day of the World of
Commodore show in Pasadena, and today was definitely
more impressive than yesterday.

    The definite highlight of today (even over seeing
Kiki Stockhammer in person and various other celebrities :)
was Louis Eggebrecht's, VP (not president like I said
yesterday) of engineering, keynote talk.  The talk focused
on the engineering aspects of Commodore.  There are
about a dozen groups in Commodore that do the engineering.
Among them are the VLSI group (the largest, I think),
the systems group, the system architecture group
(Eggebrecht's thing), CATS, the machining group --
Eggebrecht mentioned that some toolings of plastic
parts took longer than the VLSI design stage of the
Amiga -- and the product assurance group.  All in
all, they number over 200 people.  Anyway, on to
the interesting part of the talk.  BTW, the talk
was presented on an Amiga 4000 with AVPro (AmigaVision
3.0), and the speed of the display was pretty quick.
It looked like it was an 8-bit mode that was fairly
high-resolution.

    Eggebrecht first talked about their improved
design facilities, about how they could send their
designs to be manufactured elsewhere.  He mentioned
names like HP and VLSI.  For example, the Lisa
chip which replaces the Denise chip in the AGA
chipset is sourced from three different places.
He said that this has several advantages, the
new chips are done in CMOS instead of NMOS, and
designs could be turned around and tested faster.
It seems that the only thing they're using the
CSG (Commodore Semiconductor Group?) for now is
for making 6-micron parts for C-64's, which are
selling on the order of 700K/year in Europe.  That
surprised me.

    Eggebrecht, who worked at IBM as a hardware guy
before coming to Commodore, presented the future
hardware goals of Commodore first.  First off was
the low-end chipset.  He said it would be 2 chips
with about 100K transistors.  As a comparison, the
ECS uses 60K transistors and the AGA uses 80K
transistors.  It would address 32-bit DRAM and have
a 57MHz pixel clock.  It would be very much ECS and
AGA compatible.  Eggebrecht said that backwards
compatibility was very important for the low-end
stuff.  You could have a 4MB floppy drive (he said
"four megaBIT", but I think he meant byte), and
the chipset would support 32-bit processors.
The memory bandwidth would be 8x what it is now,
and 72Hz refresh rates are possible in some modes.
He said it would be "easy" to do 2x the blitter
performance and a 800x600x8 bit 72 Hz modes (!).
The low-end chip set would support 16-bit truecolor,
meaning you could have 65536 colors at once, and
a FIFO'ed serial port would be part of it.  The
FIFO will clear up a lot of present serial port
problems present now.   Gee, I wouldn't mind
having the low-end stuff now :).

    Now, Eggebrecht continued, let's talk about the
exciting stuff -- the high-end chipset.  The high-end
chipset is made of 4 chips with a total of 750K to
1M transistors.  It could use either 32-bit or 64-bit
VRAM, giving it much increased performance.  It would
have a reprogrammable pixel clock of 57 to 114 MHz.
The present pixel clock is 28MHz (AGA, I assume).  Let
me go off on a tangent here.  Apparently, Commodore's
new philosophy on systems design is to make subsystems
clock-asynchronous, so each part can be tuned to its
most optimal speed.  The decoupling of the graphics
chipset from the processor clock is one example.  Future
processors will also have asynchronous clocks from
the rest of the system.

    Anyway, going back to the new chipset, it would also
support chunky modes, without losing its bitplane nature.
It would support CD-ROM in the form of some kind of really
fast serial line where it could display video and play
back sound in realtime.  It would have framegrabbing
builtin (frame rate unknown), and screen promotion builtin.
It would have a 1Kx1K screen (I don't know if this is the
default or not).  It would also have 8 channels of 100KHz
16-bit sound (sounds like overkill, no?).  Supporting all
this would be on-demand DMA -- I assume this means we
aren't tied to having certain DMA channels for certain
parts only.  It would be backwards compatible with the
ECS and AGA, although Eggebrecht didn't stress it as much.

    The performance of this chipset is 12-20x current bandwidth.
There will be a 32-bit blitter with 8x the performance of the
present one.  The system is capable of having 1 blitter for
EVERY bitplane!  There will be 24-bit truecolor (yeah, now
where's my 48-bit color?).  There will be some kind of
enhanced decompression stuff builtin to handle different
data types.  I presume he means MPEG and JPEG.  Let's see,
what else?  The genlocking would be improved (no specifics).
The system is modular so you can upgrade to later chipsets.
The default screen refresh is 72Hz with the system capable
of 100Hz.  The processor slot would be 32-bit and processor
independent.  He mentioned RISC processors, but said
nothing more.  Wow is all I can say at this point.  But,
he wasn't through yet.

    Eggebrecht next went on to talk about the upcoming
product features.  AGA will be incorporated into all
upcoming Amiga products.  All future Amiga computers
would be modular allowing you to add stuff like RISC
processors (he said it again!).  Commodore realizes
that in the past they've kind of stayed apart from
everyone else and haven't been keeping up.  No more
of that now -- Commodore will keep up with the industry,
which means, Eggebrecht says, they will introduce 100MHz
processors as soon as they are out.  Commodore "won't
dilly-dally" anymore.  DSP support is coming.  32-bit
SCSI-II is coming around Jan 1993, and may be incorporated
into future motherboards.  CDTV is going to merge with
the Amiga line.  We're going to get full-motion video
with MPEG.  CDTV itself is going to be cost-reduced and
enhanced.  Eggebrecht stated in no uncertain terms that
Commodore still thinks CDTV is one of its most important
projects.  Software is going to be upgraded to handle all
this.

    Speaking of software, Eggebrecht also mentioned the
new OS's that are going to be coming out.  He stated that
both compatibility and stability are their two greatest
goals.  Retargetability is also a major goal (he had about
5).  Specifically, release 3 has all of the 2.1 stuff
plus AGA support and multimedia support.  3.1 has API
network extensions.  I asked Allen Havemose about this,
and he seemed to imply that the 2.0 TCP/IP software is
not going to be released but instead integrated into
the OS.  3.1 also will have file and printer sharing,
and DSP support.  Release 4 will be retargetable and
have full Postscript printer support.  I asked Havemose
about resource tracking, and he said, "Not in the next
version", meaning 4.0, but he said it will be in later.
I'm somewhat surprised at this and I hope I heard wrong.

    As to how far along these things are, Eggebrecht
says they're close to working silicon, but would not
give any dates at all.  He didn't say much about the
OS release dates.  Eggebrecht and Jim Dionne said in
very certain terms that the current Amigas (the 500,
2000, 3000) cannot be upgraded to AGA.  Eggebrecht
gave some technical reasons for it.  They include:
surface mounting of chips to reduce capacitance
when the chips are run at a high clock rate, 32-bit
access to memory, and the unsuitability of current
Chip RAM to be used.  However, Dionne said that
"about a million" people have already asked him about
a power-up deal, and there will probably be one.

    The question-and-answer session wasn't as exciting
as the talk because a whole bunch of people started
asking irrelevent questions like, "Why isn't C=
promoting C-64's and C-128's in schools?"  "Is the
C-64 and C-128 going to be discontinued", etc, etc.
It was very annoying especially since we didn't have
much time to ask Real questions in the first place.

    So all I have to say is, way to go Commodore!
You guys are doing great things and I hope things
don't get stopped as abruptly as they have been in
the past.  I think I'll be thinking hard about
powering-up to an A4000 or whatever new computer
they come out with.

--Andre
(who isn't making any of this up, really)

--
Andre Yew                             [e-mail] (131.215.48.100)