Contents | < Browse | Browse >
/// Portal Online Conference: Dave Haynie, Commodore Engineer
This is an edited transcript of a special *live* chat conference, held
on the *Portal System in the Amiga Zone chat area, on the evening
of November 7, 1993. This transcript is Copyright 1993, The Amiga Zone,
and may be freely distributed provided no further editing of any
kind is performed on it. User Groups may reprint this transcript
in their newsletters and it may be posted on Bulletin Boards, Usenet,
and FTP sites.
Our special guest speaker was Mr. Dave "Hazy" Haynie, of Commodore
West Chester. The conference took the form of a question & answer
session with Dave.
This conference took almost exactly four hours in real time. Much
superfluous yakking and hellos and g'byes have been edited out to
make for a cleaner read. Thanks to Brian J. Cerveny for his
yeoman's job of editing this capture of the conference. Thanks to
Bill Seymour for maintaining the question queue during the evening.
And thanks to all who participated. And special thanks to Dave Haynie
for his vast storehouse of Amiga knowledge and willingness to give
an entire evening to this endeavor.
The edited transcript begins here. The host was Harv Laser, head
Moderator/Sysop of The Amiga Zone.
Harv: AND NOW, IT'S MY EXTREME PLEASURE TO INTRODUCE A TEN YEAR VETERAN OF
COMMODORE ENGINEERING A MAN WHOSE NAME IS ETCHED UPON THE MOTHERBOARDS
OF YOUR COMPUTER.
____ _ _ _
| _ __ ___ _____ | | | | __ _ _ _ _ __ (_) ___
| | | |/ _` / / _ | |_| |/ _` | | | | '_ | |/ _
| |_| | (_| | V / __/ | _ | (_| | |_| | | | | | __/
|____/ __,_| _/ ___| |_| |_|__,_|__, |_| |_|_|___|
DAVE - ANY INTRO MATERIAL YOU'D CARE TO START WITH, BEFORE QUESTIONS
START... BY ALL MEANS, FIRE AWAY
At home, drawing pictures, of mountaintops...
Whoops, I thought I was Eddie Vedder for a sec, I'm OK now.
Kerry - Stratford: I have an accelerated 2000 and a Picasso II board on order.
I understand that with the new 3.1 ROMS I will be able to
use many AGA programs in 256 colors with Picasso's
Retargetable Graphics. How do I get these ROMS and
can you confirm the above. Also will they allow me to
finally upgrade my CDTV so I can run 2.0 programs and
hazy: I don't know the exact status of 3.1. Last I heard, it was, like 2.04
before it, planned as an upgrade for all systems. This would be for
A3000, A500/A2000, and probably the new ones, since it is a different
ROM. I don't know if you necessarily want to upgrade a CDTV, since
there are a number of programs that seem to be 1.3-specific.
The CD32 has what's essentially a per-title patch library to handle CDTV
titles, I don't know if they plan this sort of upgrade for CDTV or not.
Kerry - Stratford: Sorry, I use the CDTV for my kids as a computer and you
can't with the CD-32.
hazy: Like I said, I don't know of the CDTV plans.
ScottJ: Dave, has CBM decided which RISC chip to use in future Amigas? If they
haven't, which one would you prefer to see used, PowerPC, Alpha, MIPS,
hazy: We have looked at all of the RISC chips. No decision has been publicly
announced, however. Keep in mind that chances are, the chip we pick will
need to be servicable at all levels of computing, from our high enough
down through the games level. However, I don't imagine we would be going
with a chip _not_ on the list you have given, if that's enough of a clue
for you (it's all you're gonna get).
ScottJ: I hope its not SPARC, that chip's a dog. Thanks, Dave.
hazy: The next generation high-end systems will use a CPU independent local
bus, so the RISC decision isn't particularly critical at this point.
Timeus: Will there be a display enhancer for the A4000 similar to the A2000's?
hazy: The planned display enhancer for the A4000 has been tabled at present.
That means that no one is currently working on it, not that they never
will. I believe the design was offered to 3rd parties, though I don't
know if any of them picked it up or not. I'm not certain it's much
advantage over some of the existing 3rd party display cards.
soft-logik: Dave, do you have any guess as to who bought a lot of Commodore
stock last Thursday?
hazy: Rumors abound. The rumor mill has been quite active lately, with talk of
some financial improvements and some possible takeovers. However, I
don't know what's going on.
soft-logik: Well, good luck to ya.
hazy: In fact, one of best sources for rumors couldn't make our Friday lunch
this past week, so I'm at least a week behind on the best gossip.
Bjarian: Hazy, thanks for your inputs. My question is about UNIX. The Amiga
was first in V.4 and beat even SUN. It is the only viable PC platform.
What is being done now for Amiga UNIX?
hazy: Currently, nothing is being done on UNIX. Sorry, I liked it too. It
doesn't seem that Commodore management was willing to support it, though.
hazy: That has nothing to do with the troubles of the past year, it was a much
earlier decision, something nobody in Engineering was happy about. But
that does happen sometimes.
mykes: Is there any truth to the rumor that Dave's currently busy designing
the dream Amiga?
hazy: Basically, yes. I'm currently working on several things, all of which
involve the AAA chips to some extent. But of course, I don't design the
Amiga chips, I design the system they go into.
After the A3000 went out, I started thinking about What Next, as you
might imagine. A bit over two years ago, I formally started writing up a
new system architecture. We are NOW starting to implement that
architecture. If all goes well, that will be the first AAA machine.
DonM: I know you aren't/weren't marketing. Maybe you have a feel, though
for how things are going. Ex: I just got a ToysRUs catalog.
as they look to me? They GOTTA start marketing these things
whatever fiduciary finagling goes on. ga
Harv: [Note - Commodore has SAID (at WOCA) they will be introducing CD32 to
the USA as of Consumer Electronics Show in Vegas in january]
hazy: The US market has grown so small relative to the rest of the world, I
think they pretty much take any sales as good, as long as they don't cost
I have, as Harv said, heard they plan to introduce CD32 in January here,
but that they didn't have the quantities available to hit every market
for Christmas '93. However, this is apparently not a repeat of last
year, when we simply couldn't make enough systems due to supply
problems. As far as I know, the factory in the Phillipines has been
running CD32s at full force on 3 of the 4 production lines. That's full
production for a Christmas rush.
Harv: Dave - quick Q from me... tell us about the Philippines plant, if you
can. what's it making and how many?
hazy: I don't really know numbers. It's set up for surface mount, which is
what everything is these days. They have four production lines. The
fourth has been running A4000s and I suppose, anything else they needed,
over the Christmas rush time (it's hard to get away from the low end at
all this time of the year).
Aurelius: Hi Dave, Which of your current projects is the most exciting to work
on? the one you can talk about anyways?
hazy: The next generation system project. Everyone's heard about AAA, of
course, and I did build a AAA prototype machine, which we have running in
the labs of course. However, while that's fun, it's not as cool as when
I get to build My Own Thing. That will be the system that "houses" the
Aurelius: does AAA include PAULA upgrade/replacement?
hazy: The AAA system has a thing called "Mary", which is kinda-sorta Paula on
steriods. And then some.
David-L: Was the A4091 turned over to a third party? If so, who?...
hazy: Yes, the A4091 has been turned over to a third party. Management has for
some time been trying to get out of the peripherals biz, even while
acknowledging that sometimes, we have to make them. I don't know if the
third party involved has been officially announced, but having used some
of their stuff and all over the years, I think they're a good company.
David-L: Do you know where I go for support?
hazy: I don't know how the support end is managed, though I don't suppose
they're taking over the software anytime soon, so ultimately C=
Engineering will still be dealing with problems at the technical level.
I really don't know what happens at the user level.
David-L: What's Doin w/ DiskSalv - I sent my money - should I expect somethin?
hazy: As for DiskSalv, I spent all last week stuffing envelopes. Anyone who
registered at any level should get something in the mail very soon. If
you registered for updates and/or the commercial version, you'll get the
V11.28 update disk too.
colins: Dave, it seems that the PCI bus has been embraced by much of the
computer industry.What chance is there that CBM might incorporate the
PCI bus in a future Amiga? Also,will AmigaOS survive the transition to
a RISC based Amiga?
hazy: There is a very good chance Commodore will endorse the PCI bus sometime
in the future. No official announcement has been made yet, of course.
However, PCI is very much a solution to a problem I started working on
over two years ago. It's very close, in fact, to my solution to this
problem. And standard, as well. Draw your own conclusions :-)
fredness: Are there plans for an AGA Amiga with SCSI-II on the motherboard?
hazy: The A4000T (T=floorstanding) is an AGA machine with SCSI-2 on the
motherboard. That's the same NC53C710 SCSI-2 you get on the A4091, only
integrated into the motherboard. This machine isn't out yet, I don't
know the status of it, but the samples seem fairly complete. This is a
Greg Berlin project at present. No other AGA/SCSI-2 system has been
CarmenR: Hazy: Do the RISC chips [or DSP if you're gonna use them for that
matter] handle multi-tasking as elegantly as the Motorola chips do?
hazy: RISC chips are generally as good at multitasking as the Motorola chips.
Some have a slightly greater task-switch overhead, but that's really no
big deal if you're going that much faster to begin with. They vary in
their abilities at other things. Some handle interrupts as well, others
used a more primitive interrupt model. Same goes with exceptions, MMU
table walks, etc.
CarmenR: Is IDE here to stay? :) -=> GA
hazy: Industry-wise, IDE isn't going away anytime soon. Even though it
presently doesn't go much beyond 4MB/s, there are mental patients out
there building IDE controllers for VL-Bus and PCI. And the PC industry
is planning an enhanced IDE that supports four channels, more
addressing, and greater speeds. As for Commodore... I believe it'll
always make sense for the low-end systems. I don't like it for high end
systems, and I think we have taken some steps in the next generation
architecture to bring the cost of SCSI down relative to IDE. But we'll
see -- if it's free, and there's a low cost SCSI option, you shouldn't
complain too loudly.
Harv: Dave - Any comment on Colins' "Will Amiga OS survive transition to a RISC
hazy: Oh yeah. I believe it has to. The reasoning -- low end systems. You
can argue in favor of WindowNT or UNIX for high-end systems. And in fact,
I think both of those are good options, especially something like NT,
which should essentially be shrink-wrapped for any CPU supported.
However, no one's going to see any sense in outfitting a $500 RISC low-
end or CD machine with the $1200 worth of RAM, Hard Disk, etc. necessary
just to boot one of those.
Also, there are rather dubious Multimedia characteristics in those OSs.
Most of what people call "Multimedia" are a set of realtime problems.
The industry still hasn't figured out that GUI is a realtime problem. I
don't know how long it's going to take them to get the Multimedia thing
figured out properly.
colins: That was my worry Dave, something like Scala won't look too hot on
Pjotr: Let me describe a scenario: The new line of computers CBM will
concentrate on. One main RISC architecture, Windows NT (f ex?),
Multimedia libraries (incl video) as primary advantage over NT PCs etc,
Interfaces for Video CD cable, PCI(?), Video conferencing standards etc.
Q: How will the next Amiga stand out without being propriery?
hazy: On the hardware side, you'll still have Amiga chips. There are things in
AAA which will make a very impressive difference in many of the things
we're doing with computers these days. But hardware is expensive, and
time consuming to design. If we have to build every piece of the system
from scratch, it's going to take longer and cost more than comparable
"standard architecture" systems. Everyone has pretty much figured this
out, except maybe Sun. DEC and Apple certainly have. I think you have
to concentrate on the things that will make your system stand out.
On the software side, like I said, I think we'll still have AmigaOS
around. Check out the last BYTE for their preceived effect of NT or
OS/2 on your hardware, relative to Windows.
From C='s point of view, supporting NT isn't a big deal if the shrinkwrap
version is available for your processor, so why not? Yet I think the
processors (RISC-based) need to run Amiga binaries to make the system an
Amiga straight-off. Of course, keep in mind I'm not the management or
the software group, but that's the way I see it.
Furr: I would like clarification on the Z3 BUSTER DMA mess; I've
gotten conflicting stories from different people. I have an
A3000, and I want to know if there is a new BUSTER available
for this machine that fixes the Zorro 3 DMA problems. I am
considering a Z3 board that this is apparently a consideration
for (the GVP Spectrum/24).
hazy: The Buster chip in the A3000 (Rev G or -07) does not support Zorro III
DMA. That feature was simply left out, to get the machine out on time.
You can get the latest, Rev K or -11, which supports Zorro III DMA and
also fixes a bug in the Zorro II DMA to Chip RAM on the A3000.
I suppose in the USA, you would have to order this part through your
dealer or one of the C= parts specialty places. I don't know who's in
charge of replacement parts anymore, it might be SMG.
Furr: Also, when might the first machines based around the AAA
chipset actually reach market, and will those first AAA machines
still use 680x0 processors, or the as-yet undecided RISC chip?
hazy: The first AAA machines will ship with 680x0 processors, hopefully the
68040 and 68060 will both be options, assuming Motorola's silicon
development keeps paces with ours. For RISC, you would simply plug in
a RISC CPU module.
KyleW: Is it true that AAA will not work on AGA machines (my UG refuses to
upgrade it's antique A500 because of that rumor) and will AAA be
available for older machines (I have way too much money in my 2500 to
upgrade any time soon).
hazy: The AAA chip set is a radical departure from anything you have seen
before. Everything in it was designed new, from the ground up. There
is no way possible to retrofit it into any older system other than
building a Zorro III card for A3000/A4000 class systems. That certainly
could be done, it's up to management to order this or not. It's not
practical to implement AAA on a 16-bit card (I won't say impossible, but
it would be a great deal of trouble).
LadyHawke: Hi Dave, nice to see you again, we owe you at least a cement-mixer
full of macadamia nuts by now ...
hazy: That's a rather curious image.
LadyHawke: My question is more of a dilemma than a question. I have an A3000.
The A4000T at the show looked good, more slots, two hard drive
Harv: [and two video slots!]
LadyHawke: I am at a loss as to whether to continue to upgrade my A3000, since
every dollar spent on that is a dollar less to save for a new
system, and now we have even further developments. I'm worried
that there is less support for the 3000 than I expected, and then
RISC machines on the way. What would you do if you were a high end
consumer? (and computer artist)
hazy: Well, part of the question is, just what do you want for the A3000?
Barring a DMA upgrade, Zorro is Zorro. AGA is a big improvement for
some, maybe especially if you're into art.
I still have an A3000 here. It's got an '040 card, a 3rd party display
card, ab18MB and about a gig of hard disk space. I'll admit the A4000T
is a bit tempting, but I don't know (I didn't work on it, so I don't
necessarily get a freebie or anything).
Harv: [a friend of mine has an 040 3000 with 84 meg of ram.. I hate him :-) ]
hazy: It's always going to come down to new stuff coming along. The very
nature of this business is Faster, Bigger, More, ASAP.
I have a long-term goal to make systems far more modular than they have
been, and we just might get that in the next generation. It hasn't been
technically feasible before.
LadyHawke: Dave, how long before another system comes along?
hazy: Lew says AAA will hit sometime in '94. Don't expect it for another
year, though, to be reasonable about it. I'll guarantee you someone
buys an A4000 the day we announce the new machines. These things
Jim-Guy: My question is about using my Video Toaster on the A4000 with a 1960
monitor.. C= was supposed to build an adapter? Has this been built..
will I ever be able to use my 1960 on my A4000 so I get full AGA
hazy: I have seen these around the labs. Apparently the first batch came in
from soemme cave in China marginally functional. They do exist. I
think they're probably just in short supply at present.
Jim-Guy: What is the relationship or how is the relationship between CBM and
hazy: We seem to get along fine with NewTek these days, at least from what I
can tell. Personally, I have a blast every time I get together with
Teletran: Dave, What's the general feeling at C= in regards to how well the
CD^32 has doneso far (acceptance, sales, etc.)?...
ALso, any thoughts (personal and/or professional) regarding 3-DO?
hazy: For whatever reasons, we're often the last to know about sales. But the
word on the street, or by the coffee machine, is that CD32 is doing well.
I will admit to being a real skeptic during its development. But I think
it was done very well on both the software are hardware levels. And they
kept the proice pretty reasonable.
Also, MPEG is as cool as Jeff Porter originally claimed it would be.
I think you have to give the coolness edge to 3DO, of course. CD32 is a
big upgrade to existing games machines, but 3DO could be out of this
world. Then again, what do expect me to say -- I have all the respect in
the world for Dave and R.J., and every day more of my friends work for
3DO :( They're aren't a guaranteed success yet, though. Remember back
when CDTV came out and everyone hard that zillions of companies would be
making Phillips CD-I players. Seems I can still count CD-I players on
one finger. 3DO will certainly do better than this, since it's going
into lots of places other than "game machine"
The price point is a problem now, no other CD machine has been successful
up there, at least in the volumes they were after. They'll need
3DO-specific games. They will have some, but they'll also have ports of
PC games. They may have to count on stuff filtering down from embedded
arcade machines before they really differentiate themselves. The SGI
technology is very cool, at least on a $10,000+ workstation.
What you'll get in the home for $250 is another story. Nintendo may be
able to play the old game of CPU power vs. cool display hardware to some
extent. But if they wind up with an embedded R3000-derivative rather
than a semi-decent R4x00, they won't be much beyond the 3DO's CPU.
lbperez: Greetings...just wondering when are we going to see the expansion
box for the CD 32.
hazy: I haven't heard any release dates on it. It's real, though, a number of
the low-end guys asked me about various bits and pieces for it.
And I suppose you may have noticed, if you have been pricing CD-ROM
drives, that the CD-32 is about the same price. I'm certain interface
solutions will be quick in coming.
wms: Last time I heard you in conference here on Portal, was after some
bloodletting at C= and ewhac asked "Is the light still shining,
so to speak?" You replied that you were withholding judgement
for the moment. How do you feel now? And how is morale in general?
hazy: Well, it's a tough call. There are good things happening, no doubts.
AAA is progressing, management has made committments for new chip
revisions on an actual schedule, etc. CD32 seems to be doing what A1200
was supposed to have done last Christmas -- make some money. And I am
getting to work on the architecture I have been designing for the last
two years. On the other hand, there has been lots of attrition in
I expected this, having been through exactly this same kind of thing
back in '85-'86, but it's never something to build morale. While it's
impossible to say, I believe a great deal of this could have been
prevented with some good management. As well as morale in general.
Things may even be turning on these points, too, though perhaps it's too
early to tell. We are getting some new hired in HW to replace a few who
have left. I imagine SW is too, though they lost more key people than
elc: Motorola driven AGA is slow...will Motorola driven AAA be faster? Say the
Indy were to drop in price why should my next computer purchase be an
Amiga? What is it really ofering that I can't get anywhere else...
Harv: [go price an Indy. if you want 24 bit and a hard drive I hope you have
hazy: AGA is, in general, slow. Slow, certainly, as compared to what you
expect in modern 32-bit systems. There's a good reason for this -- it's
based largely on the ECS architecture.
elc: The Amiga always seems to be just shy of what the Amiga community is
looking for, will the next generation Amiga wait for "perfection"?
hazy: AAA is completely new. While I don't expect it to be the fastest system
known to man for all things, it will be a substantial improvement.
For instance, a demo in the lab is blitting around 24-bit images
substantially faster than AA can move an 8-bit image.
It has lots of good features, compressed-video modes to speed up
animations, copper-programmed-blitter to help offload the CPU, etc.
The current system architecture plans call for additional system support
to speed things like RAM to Chip RAM transfers. You can have either 8MB
or 16MB of Chip RAM, depending on system configuration. A VRAM system
consumes no chip bus bandwidth for video fetch. Lots of good stuff in
I don't know how you compare it to an Indy, they're different systems.
An R4000 is faster than any 68040, and we don't know what the '060 will
do just yet.
MarcR: Will there be a Laptop Amiga? Will there be more "specialized" Amigas
like CD32 but for professional and other uses? Will AAA bout out too
hazy: They have no plans for a laptop Amiga at present. I want one too, I
think it may take a 3rd party to actually get one (keep in mind most 3rd
party video displays run on a VGA of some kind).
There will definitely be more "specialty" Amigas. I don't work on them,
but we do have a group focusing on that kind of thing. I suppose if they
went high-end, I might even get involved. One big advantage of the next
generation high-end architecture is that it's designed to implement
amodular systems. Most previous architectures did one thing well, and
got in your way when you tried to do something else with them.
I don't think AAA is coming out too late, though it's later than I (and
all of you, of course) would have liked. There's still not a big move
to 4MB floppies yet, everyone's still trying to work out just how to do
multimedia, and I think AAA supports much of this in hardware, the right
MarcR: Would C= licence the Amiga chips for someone to make a laptop, etc.?
hazy: I know C= is much more interested in licensing, and does have plans for
chipset deals, at least into the markets they're after. Jeff Porter might
know more about this.
JWolf: What are all the details about the AAA chipset? Any other new custom
chips? Details? Also, when is AmigaDOS 3.1 due out? What new features are
in AmigaDOS 3.1 (both under the hood and over the hood)?
hazy: Well, I would need hours to discuss all of AAA, and I'd probably have to
read the manuals.
I can tell you it consists of four custom chips: Andrea, Mary, Monica,
and Linda. They are all full 32-bit chips, and actually, Linda and
Monica have a 64-bit mode. The chips support 8 MB of Chip RAM in
32-bit mode, 16MB of Chip RAM in 64-bit mode. RAM can be DRAM or VRAM,
and can actually be mixed in a system on 1MB/2MB boundaries.
Andrea is the Agnus/Alice replacement. It supports the old 16-bit
registers, new 32-bit registers, enhanced Blitter and Copper, Burst mode
to Chip RAM, and display rates up to 110MHz.
Mary is the Paula replacement. It has 8 16-bit audio channels, an
enhanced flooppy interface that can handle 4MB floppies, 150KB/s CD-ROM,
and probably even ST-506 hard disks if anyone cared to dust one off and
Linda is a smart Line Buffer chip. It takes data from Chip RAM, as
directed by Andrea, and assembles it into a scan line. This allows bursts
from Chip RAM to go long, and allows the chip bus rate to be decoupled
from the pixel speed.
CD-ROM support was kind of a freebie. The Mary chip has a few data
formats, as low-level data formats, one of which is the CD-ROM format.
While we expect 4x CD-ROMs soon, and CD-ROM on SCSI as the
standard way in a high-end machine, the CD-ROM format throught the floppy
interface does illustrate how fast the floppy interface is now.
Monica is the Denise/Lisa replacement. This handles a variety of planar
and chiunky display modes, HAM and some new compressed mode included. It
can handle all kinds of resolutions, and with the variable pixel clock,
they can be just about anything you want, assuming the system can
provide that clock.
The chip designers name the chips.
NES-Bill: Serial ports Dave?
hazy: Two buffered serial ports.
DeckApe: 1) Has the A4000T hit the stores yet? If not, when is it expected to?
What is its list price....
hazy: Far as I know, the A4000T has not shipped. I don't know any details on
DeckApe: 2) Will the A4000T be outdated technology within 24 months of it's
hazy: Will it be outdated? Well, on some level, anything that ships is
outdated, since work on its replacement is already underway.
It certainly won't be useless. A machine remains what it was when you
bought it. However, better stuff will always come along.
Harv: [lots of people still use Amiga 1000s. I bet some people here are using
them right now]
DeckApe: [Won't bet - *I* am one of the 1000 users!]
hazy: My dad bought a C= PC-10 III system maybe 4-5 years ago. That's a 12MHz
8088 or some-such. At a computer faire today, I saw a stack of 8088
machines going for $25 each. He paid around $500, C= discount at the
time. You have to decide when it's right to upgrade and when it isn't,
and I can't tell you what your needs are.
Aaron: With whatever RISC Chip that C= Chooses/chosen (PowerPC I hope)
will they be doing the same as Apple with Emulation Q. This
Would allow the software to run in 68k Mode but if it is
Compliled for the PowerPc it can run in PowerPC Native mode
Giving it a the full power of the very powerful Power PC chip.
hazy: Again, I'm not the software group, but I anticipate a 680x0 emulator
would go into any RISC OS we ship. It's really not that difficult, in
that only user-mode stuff has to be emulated.
They can play tricks with the OS, using the emulator or native code as
they see fit. And of course, the underlying hardware is the same, so you
don't have to emulate it. That's perhaps the worst thing a PC Emulator
has to do.
Aaron: Second Question: How expensive is it to make the AGA chipset
Compared to the Future AA+ and AAA. If the AA+ or the AAA is
the same or just a bit more will C= put them in the CD32.
Would this not bring the CD32 power up past the 3DO's??
hazy: As far as chip prices, AGA is chip. Real chip. Like, cheaper than ECS,
as I understand it, since it yields much better.
AAA will cost more, though much less than we originally guessed, thanks
to the downsizing of chips (AAA takes about a million transistors).
A AAA game machine is certainly possible, though I'm not sure if it's the
AAA has a lot of features than make more sense in a computer than a game
machine. I mean, 3DO certainly didn't waste any silicon making sure you
could get 1280x1024 noninterlaced.
AmiGadget: Dave, it's obvious CBM is convinced, that there are plenty TOMORROWS
for Risc Amigas, AAA chipsies 4000-T's, etc. This stuff is two years
late, now. By the time you get it engineered, there'll be ZERO
Amiga software developers with any capital to support it. With all
due respect for you personally and NO respect for C=, I see no
sense of urgency, yours or CBM's, to DO something NOW. DO you ...
think there's plenty of time? Or should we all just set our clocks
for 2099, when that year's LAST year's technology will then match
what we can buy on several other platforms now.?
Harv: [for you IRC people... AmiGadget publishes a magazine by the same name]
hazy: Well, I can't make things happen any faster than I'm making them happen.
I wish they were moving faster, but we're doing what we can. I think
you'll find this new system isn't lacking as a '94 machine.
I can't get into too much detail, only to say that [a] I have been
working on this architecture for two years, something that never happened
at C= before (two weeks was more like it in the...
old days), and [b] I'm doing things with system throughput no one else is
doing, or at least talking about, now. No big surprise, really, though.
There's little incentive to get too fancy in "standard" architecture.
Apple doesn't build clever hardware, and they seem to be wanting out of
the hardware business anyway.
AmiGadget: ALSO... you keep saying "high end"- what kinda money? $12k, $4k?
hazy: C= "high-end" is $4k and below, generally. I think it should start at
about $1.5K-$2K, less if possible.
High end is the stuff that's not game oriented. "That which does not
suck" one might say. I don't think we're going for $10K systems ever,
there's no market there.
mykes: Has C= considered distributed computing? ...
hazy: We have done some distributed computing. Nothing real fancy, but
It started out with a renderer and a custom CPU server that allowed one
machine to request CPU time of anyone participating over the network.
This wasn't formalized, but it's a good model, and could be formalized
over Envoy without too much trouble. There has also been some work with a
distributed database model, though I don't know much about that one.
mykes: Which 3rd party gfx board do you own, if any? ...
hazy: I have a thing called oMniBus, designed by a guy named Oliver Bausch
in Germany and marketed through ArMax. This is a passive bridge card that
puts Workbench up on the VGA card of your choice.
It's cool, it works, it's faster at many things than AGA, does 256 colors
(true color with a special loader program),
and works under 2.x and 3.x. This never made the US market, unfortunatel
y, and probably scared off Bridge Card users. But it does me just fine,
at up to 1280x1024.
mykes: ISA is faster than ECS...
hazy: Actually, the ISA/Zorro II CPU to RAM interface is 1/2 that of the CPU to
AGA interface speed. But the VGA card never blocks CPU access.
mykes: Do you feel that C= will be able to crawl out of the niche market
and into mainstream? ...
hazy: Well, at least we can pursue the niches. I would like to see us grow
and hit the mainstream, at least some mainstream. But increasingly, the
mainstream is defined by the the software you run,
it has NOTHING to do with your hardware platform. Or, at least, it
shouldn't, and won't if MicroSoft has their way. What can you do against
that, I don't know.
[a] SGI only sells in niche markets [b] SGI machines run Windows NT, a
mykes: When's the new R.E.M. album coming out? And do you have it on your
workbench along with your other prerelease stuff? :-)
hazy: Last summer, REM said the new album would be out this fall and they'd
do a tour. Now they say the album will be out maybe in the winter and
they might not tour before 2000.
mykes: most important question of all :-)
Furr: As far as the IDE/SCSI business goes, there are now some MS-DOS
systems that use IDE for the primary hard drive, and have a chip
socket and connector on the motherboard for SCSI, if the user
wants it. I think this would be an excellent route to go, in that
it gives you an inexpensive HD interface, and no-slot SCSI access
for those who need it; what do you think?
hazy: Well, the high performance SCSI chip we use now is a 160 pin PQFP
package, it doesn't exactly drop into a socket (well, anything like a
production-worthy socket). I do agree with the spirit of this --
If you do go IDE, SCSI should be a simple addition, it should not require
a full expansion card with all the inherent costs of such a card. My
personal opinion, of course.
Furr: Now that we're getting one-chip Ethernet interfaces, how long
do you think it will be before a 10baseT port is standard on the
Amiga motherboard? Artisoft [maker of the LanTastic networking
software for the MS-DOS world, for those who don't know] is
already calling their ALICE chip "The Serial Port of the 90's."
hazy: Actually, most of the UNIX and all the networking magazines seem to be
pushing 10-Base-T as a kind of serial port-ish solution. Which is
appropriate, since you can't really use it directly to network anything.
However, given that 10-Base-T chips are getting motherboard-cheap, and
trancrivers over to 10-Base-2 are also cheap (as low as $50), I would
definitely support 10-Base-T as the most practical networking solution
Furr: Are there any new BridgeCards or other similar emulation items
coming from Commodore, or will that market be left to 3rd parties?
hazy: There has been nothing beyond the '386SX BridgeCard, at least
discussed in public.
Furr: What is your personal opinion of the Commodore Stockholder's Movement?
hazy: As a stockholder, of course I'm concerned about the relatively minor
investment I have in C= stock. As a C= employee, I'm far more concerned
about the long term viability of C=. Since it seems to me that the
Commodore Stockholder's Movement has only the Amiga and, as a result,
Commodore's best interests in mind, I couldn't rightly do anything but
applaud the efforts.
RMills: I work at Warner bros. animation and see plenty of SGI and Macs come
in for testing and wondered who or if Commodore would do this?...who
should I get in contact with there? Getting my A4000 in my office
was like pulling teeth.
hazy: Technically, that kind of request should go through Marketing. But I
don't know who would be in charge of that.
RMills: Can you give me a name?
hazy: I would recommend talking to Jeff Porter at Commodore, in Engineering.
RMills: ok I really think stations set up at Warners could be a great plug for
Commodore and they seem to miss this as a viable thing for them.
hazy: Jeff does a fair share of Evangelizing in addition to running what's
basically the "multimedia" branch of engineering.
Harv: [Jeff P did a good job with his speeches at WOCA in pasadena]
hazy: Well, the US does have new Marketing Prez. If his hands aren't tied too
much, maybe he's interested in doing something.
Timeus: Dave, How hard would it be for a developer to put external instruction
cache on an a4000 accelerator card?
hazy: You could build a cache into the design of a CPU card -- the first C=
'040 prototype had a 128K cache. However, it's probably not worthwhile.
You can just about always get more performance by going faster than
caching, at least until you run out of clock speed (eg, you can't get
'040s beyond 40MHz). We got between 5% and 15% speedup in most code
with the 128K cache. As for Apple, I assume they didn't sue Commodore
because they had insufficient legal grounds for a suit.
Timeus: So you actually looked into it. Impressive. And why did't Apple sue
Commodore over look and feel as it did to Microsoft and HP. The amiga
has the better interface and the Mac is a poor copy of Amiga.
hazy: Keep in mind that MicroSoft licensed some stuff from Apple for Windows
1.0, and since HP's NewWave is built on top of Windows, anything that can
be said for WInindows can probably be said for NewWave, at least legally.
There were plenty of elements in the Amiga's GUI that didn't enter the
Mac's until later. So it was also possible that C= would have had just as
much on Apple as they could have dug up on C=.
Of course, C= probably could have done like NeXT and paid Xerox PARC for
a GUI license, which should sufficiently stimie Apple's attempts at any
KyleW: the window/icon interface was developed by U of AZ and is PD!
RedWine: Do you forsee any conflict between the interests of RTG and the
apparent functionality of the AAA chipset? Or is it safe to
assume that anything AAA will support will also be available to
any other cards that come along from other parties?
hazy: No. In fact, some features of AAA, such as chunky pixel support, require
at least a degree of RTG to built into the system. There's a good chance
that most vanilla chunky display cards won't even need a full RTG driver
as a result, since that mechanism will be built-in for AAA.
RedWine: With respect to the slow memory system of the 4000, why was the
decision made to not include a more appropriately fast memory
capability on the 68040 card itself?
hazy: The card that's in the A4000 was designed to be the lowest cost '040 card
possible without going to custom logic. That necessitated some
compromises, but it did help on cost. Welcome to the world of the
desktop system. :-)
RedWine: What's the standing the CD-ROM drives for AGA machies, and is it
technically feasible to include the advantage of chunky->planar
hardware from the CD32's Akiko chip?
Aaron: Is the Chunk->planar Feature included in AAA and AA+?
hazy: Well, I thought about buying a CD-ROM drive for my A3000 today. The CD
FileSystem is built in to 3.1 and beyond. You don't need any special
magic to get one. The chunky-to-planar conversion (what we call
corner-turn memory) is in Akiko, it's not possible to add that as-is to
any other system. I expect my version of corner turn memory (which
predates Hedley's, though <he didn't know about it) will show up in the
RedWine: I'd really LOVE to get a 4000 keyboard hooked to my 1200. Is
this a real tough task, or is it something a college senior in
computer engineering like me could take on? :)
hazy: The A1200 is a tricky one, since the keyboard micro (normally in your
keyboard0 ) sits on the A12000 motherboard. You can't easily get at the
CIA serial port lines that connect between that micro and the CIA.
Also, like the A500, the A1200 gets its reset via a dedicated line from
the keuyyboard, it doesn't understand the encoding used on "high-end"
keyboards. So if you did hook it in, C-A-A wouldn't work.
LadyHawke: Re: CD^32, Dave. When I first saw/heard about it $400 for a game
machine felt a little consumer-pricey on thinking about it some
more, the other way around, with it as a CD-ROM play with cache,
memory, fast CPU, that is, a FREE computer atttached, it took on a
different appeal. So how feasible is it to make it into a
whiz-bang CD-ROM for multisession, multiplatform?
(with a FREE computer attached)
hazy: A few developers have already expressed interest in providing the
LadyHawke: $365 for the most awesome CD-ROM player on the market takes on a
whole different perspective
DeanF: Julie...not unless there's awesome software for it
LadyHawke: Dean, if it's multiplatform, there *is* software for it
Harv: julie - you talkin about playing ANY CD-ROM format in it? Mac? MPC? CD-I?
3DO? etc.? is that what you mean?
LadyHawke: Yup, Harv, I mean multiplatform capability/interfaces
hazy: Yeah, I mentioned earlier looking at CD-ROM drives today at the computer
fair. $175 for a Teac 150kB/s, $350 or so for a Sony 300kB/s.
I didn't buy one, but I am still in the market :-)
Harv: $150 for an NEC CDR25 from NES-Bill (see him outside :-)
Aaron: Heck $360 for double speed Multisession CD Rom drive alone isn't that
bad.. with the CD^32 you get a game machine and computer with it....
CarmenR: Hazy: Fist of all, I'm assuming that when you refer to "Next
Generation" and "Hi/Low End" and "A5000", you're talking about the
same thing. How much of the A5000 will be 32-bit? How much 64-bit
[zorro slots, etc]. Will there be any part of it higher than 64 bit?
The CPU slot perhaps?
hazy: Basically. "Next Generation" defines the architecture. "A5000" would
define one instance of that architecture. Most of the system will be
32-bit, though generally everything will have upward/downward paths to
full 64-bit implementations.
AAA register-wise is all 32-bit (except where necessary for backward
compatibility), though display fetching (the GD and VD buses, to get
technical) are either 32 or 64-bit, depending on the configuration.
The system's Fast RAM will certainly be 64-bit (yes, even feeding a 32-
bit processor), though we have looked into 128-bit versions, and may use
this if it improves things even more.
No enhancements are planned to the Zorro III architecture, though
certainly we'll have a much better implementation.
Data bus width isn't everything, throughput is what you really care
about. There's perhaps an aggregate bandwith of 400-600MB/s in a basic
next generation machine.
JimB: Will allowances be made for more than 5 slots???
hazy: We may have more slots, certainly I would expect more in a tower system.
CarmenR: And will the case be more attractive than the A4000 case [pretty
hazy: I don't know what they'll plan for a case. It's virtually certain we'll
be using more standard parts, like we did in the A4000 and even moreso in
the A4000T case (which most maybe haven't seen yet).
Harv: the 4000T case is very pretty (there's a pic of it in the library here)..
has a cool smoked plastic door in the front too.
CarmenR: Dave: I've seen the A4000T case.. It's decent.. I love the A3000
case.. The A4000 case bites it.
hazy: I like the A4000T case too. And keep in mind that uses more standard
parts than the A4000. They'll always do a custom bezel of some kind.
The A4000 case was a compromise -- it was supposed to be the "universal"
case, for Amigas and PCs. Only, the PC guys got in there first, then we
stopped making them.
Pjotr: Can you tell me the status and plans for the software side of AAA
development? (OS support, multimedia support etc)
hazy: Some AAA stuff will get "built-in" like with the current chips, some
may only be available via RTG drivers, and I suspect some will get really
hot under specialfx.library.
Pjotr: Workbench is TM CBM? I have seen Workbench used for other software,
why not sue?
hazy: I believe they have a TM on Workbench, but who knows. AT&T has a
"writers workbench" a zillion years ago. My great grandfather had a
workbench in his wood shop. It may he hard to defend.
Pjotr: How's your kid doing? Old enough for a CD32 now? :)
hazy: Sean is in his two's. He's old enough to demolish a CD32 in short
order, though I suppose I will be setting up something for him before
long. Brat #2 is due in March. She's the last one.
NES-Bill: Dave, last time you were we were chatting here you mentioned
some work you were doing on the side with a hack for the
4000 that would allow for more motherboard memory. Something
similar to the old 'piggyback memory' hack for the 1000.
What, if anything, has come of that? Needless to say, I
think that a lot of people would be interested in the
potential for 64M on their motherboards. :-)
hazy: Yeah, uh, umm, well, er... Seriously, though, it's still alive and well
and fermenting in my mind. No real trick, though, other than tracking
down someone who'll lend me some 8MB or 16MB SIMMs to play with. I have
been really busy with DiskSalv stuff, that's my only excuse.
Harv: THAT'S IT! DAVE WE CAN'T THANK YOU ENOUGH FOR SPENDING NEARLY FOUR HOURS
IN HERE ANSWERING ALL THESE QUESTIONS.
[many lines of "applause" and salutations deleted].
What follows is a listing of the names of everyone in attendance at
this special conference. The format is the output of the "who" command
in Portal's chatting software, which is "log in name (chat handle)".
In total, 84 people attended, although not all of them asked questions.
Aaron Mustang Smith (Aaron)
Andy - Finkel (Andy - Finkel)
Brian - Dziki (Brian - Dziki)
Bronx (Robert E Wanser)
Daniel J McCoy (DMcCoy)
Jim - Marias (Jim - Marias)
John D Short (Jodash)
Kerry - Stratford (KerryS)
Michael - Meshew (Michael - Meshew)
Thomas William Clarke (TomC.)
== end of transcript ==