back2roots/padua/padua.7z/padua/text/Haynie.interview

Internet Message Format | 2014-05-19 | 23.8 KB

Path: cs.tu-berlin.de!zrz.TU-Berlin.DE!netmbx.de!Germany.EU.net!EU.net!howland.reston.ans.net!agate!msuinfo!harbinger.cc.monash.edu.au!yarrina.connect.com.au!werple.apana.org.au!bushwire.apana.org.au!guru!Sam_Yates
From: [e-mail] (Seadog)
Subject: Haynie conference transcript
Newsgroups: comp.sys.amiga.misc
Summary:
Keywords:
X-Newsreader: TIN [version 1.1 PL9]
Message-ID: <[e-mail]>
Date: 12 Feb 94 02:45:12 +1000
Organization: Guru Meditation BBS
Lines: 537

This is a transcript of a conference with David Haynie and Randell Jesup.

---------------------------- Cut here ---------------------------------

10:02:31 PM EST Thursday, February 3, 1994

MarkM/MOD:
<bang> <bang> <bang> goes the gavel. :-)
I would like to welcome you all to this online conference.  It has been a
long time since we have had a formal CO.  I plan on getting many of them
put together in upcomming months.  Later this month we will have Utilities
Unlimited as well as GVP.  I have others slated for later.

Let me go ahead and send this formal stuff...

Greetings.
I would like to welcome our special guests Dave Haynie and Randell Jesup
from Commodore. Dave Haynie is a Senior Systems Engineer with Commodore
International Services Group.

Randell Jesup is the Operating Systems Development leader with Commodore
International Services group.

I ask that everyone keep from asking 'marketing' questions, or questions
about particular issues dealing with the sales of Amiga products.
These engineers have agreed to speak with us about current technology and
announced technology.

Since this is a formal conference, please ask questions by typing a
'?" first.  When I tell you to go ahead, then send your question.
At the end of your question--type GA to indicate the end of your
question so that the guest can answer.

If you have a follow up comment, enter a ! and I'll recognize you.

Randell Jesup:
Before we start, I'd like to let people know there
are things we can't talk about, of course.  Please keep that in mind.

KarlK: Guys, is anything being done about the port speed in the
next generation?

Randell Jesup:
Port speed? which port?

KarlK:
Serial and Par

Dave Haynie:
We realize some limitations are present in the current serial
implementation.  The main problem is the lack of a FIFO, something that was
added in the latest generation of PClone serial ports.  We have looked into
addressing that.  On the other hand, we don't have any immediate plans to
offer a complete IEEE 1284 parallel port, but it is likely that future
systems will offer a faster mode that could conceivable become an IEEE 1284
work-alike.

Ron Romine:
Does the Custom Chips (Lisa & Alice) run at 7mhz or 14mhz? And if 14mhz,
have they always had that speed capibility?

Dave Haynie:
I guess that's a HW question too.  The concept of "running at" a specific
clock speed is an architecture-dependent thing.  The Lisa chip's input clock
is 28MHz (nominal), the Alice's inp ut clock is 14MHz.  Alice runs the same
bus cycle as a 14MHz 68000, in terms of clock count.  Lisa's data transfer
is more like that of a 14MHz 68030, in terms of how much data is transferred
per cycle (eg, Lisa runs a 32-bit burstcycle). The ECS and original chip set
was also equivalent to a 14MHz 68000 in cycle speed.

Clark:
Any ideas to add a true spooler for printing & is DSP going to be there?

Randell Jesup:
No one is currently working on a spooler, though there are versions
available
in the PD I think.

As for DSP, there are already some DSP boards available (sunrise), and we've
been making projects we don't currently have time to finish are available to
developers.  I think you'll see the ex-Commodore DSP board soon.

MarkM/MOD:
What is the real problem with A4000s that will not cold boot?

Dave Haynie:
The last I heard on that problem, it was some kind of start up problem with
the Seagate IDE hard drives.  I don't know the final analysis, although I
don't believe it was a stiction problem (similar symptoms were due to this
on some Quantums a few years back).  Greg Berlin did find and address the
problem several months back.

Randell Jesup:
Also, I think it was an interaction between the power-supply rise-time and
the drive.  It wasn't stiction, I heard the results from Schilling.

Ron Romine:
Is this a problem with IDE SPT and battmem not waiting long enough?

Randell Jesup:
No, it's the rise time of the PS voltage confusing the drive.  Battmem isn't
involved.  Other drives are fine.

Dave Haynie:
The problem addressed was a drive-specific thing, related to the drive's
initialization interacting with the power-supply startup.  This is NOT the
"traditional" Seagate slow-boot problem, which is a SCSI-specific thing if
I'm not mistaken.

Dean/DKB:
In the next released OS, will there be support for multiple printers?

Randell Jesup:
Printers: If we can find time and resources, we'd like to get that done.
It shouldn't be too hard. It's not top priority, of course.

Jim Philippou:
What type & speed processor are you planning for the Amiga NG?

Dave Haynie:
The next generation (which, incidently, will try to give some consideration
to A3000/A4000 owners) will pick up with high speed '040s and '060s.  Given
a modular processor interface, and of course what you have all heard from
Lew Eggebrecht's talks, RISC is a drop-in at some point as well.

Jim Philippou:
Are you planning to build any type CPU board that fits into an A3000 with
the higher processing power?

Dave Haynie:
The A3000 currently supports A4000 processor modules (I'm soaking in one).
The real problem, and why we did not recommend the A3640 module for use in
the A3000, is that the 68040 came out significantly hotter than we had
planned for.  Going to 3V in future processors, plus some things we can do
to our own system implementation, should (eg, we believe so at this point,
I can't promise with absolute [certainty]) yield modules that work in all
A3000/A4000.

Ron Romine:
Will "3.1" be a Workbench upgrade for "3.0-Roms", or require 3.1-Roms.

Randell Jesup:
3.1 WB will not require 3.1 roms.  It will require 3.0 or above.

Mike Smith:
So when will we probably maybe see the "5000"?? i.e. is it finished yet?

Dave Haynie:
I can't give you a date for the Next Generation Machine (call it A5000 if
you like, the name is a marketing decision, of course).

I have been at work on next generation technologies for the past 2-3 years,
our chip group longer than that.  We know what the system is going to look
like, and have some idea of when things will be ready.  The design was done
a bit differently than in the past.  Things are intended to be more modular.
So you may see the first of these new things show up for the A4000 before
the next generation system(s) actually are ready.

Chris Tolmie:
When  will we see C= support CDROM drives for the A2-4000 series of Amigas?

Randell Jesup:
If you mean other than SCSI drives, I think Lew has mentioned that we're
working on some adapters.  Not so much for getting cheap CDROMs (though that
may be), but so you can use CD32 titles and access methods (and mpeg). We
already support SCSI CDROM drives, and 3.1 has a CDFileSystem (from
CDTV/CD32)
included.

Erik Flom:
Re: the "Internal Audio Conn." on A4000s, what are the pinouts for this
connector?  I've found a vendor with and adaptor, but the sound level of the
external audio is about half of what the internal Amiga Audio is.  Is there
some way of controlling the input level? (Like on PC sounds boards? :^)

Dave Haynie:
I don't have the A4000 schematics handy.  Last I recall, it was a 3-pin
header that just mixes into the traditional Amiga audio output.  We did a
similar thing on the A3000T.  This lets you hook up a CD-ROM or DSP audio
source without the need for external mixing.

Stuart H. Brand:
Re: AAA sound support, will it be 8 Ch, 16 bit?

Dave Haynie:
The planned AAA audio subsystem is essentially an upgrade of the traditional
Amiga audio.  Rather than four DMA channels at 8-bits/channel (6 volume), it
supports 8 DMA channels at 16-bits/channel (with volume, though I don't
recall the resolution).  There is some question as to whether this
traditional
Amiga solution is the best way to go, since the DSP technology we developed
(but have yet to place) offers more flexibility.  I think you can count on
16-bit audio in the next generation system, hopefully it will be the best
system for the $$$ we can provide.

Ron Romine:
Can the current A1200 IDE-HD interface support an IDE CD-ROM drive, if the
drive matched IDE standards and provided its own power supply?

Randell Jesup:
Well, it might.  I don't have an IDE CDROM.  If it acts like a disk drive,
including acting as if it's using 512-byte sectors, then it should probably
work.  It won't notice disk removals, at least in anything before 3.1
(in 3.1 it might).  However, if the startup code decides that the thing
attached is not a disk, it won't let you access it.  So the answer is:
maybe.
The ATA committee is working on something called ATAPI.  ATAPI is basically
SCSI over an AT interface.  To use it, we'd need a new driver.  I follow
both
the SCSI and ATA committees.

Jim Philippou:
What are the major features in V3.1 compaired to V2.04 and can you comment
on availability?

Randell Jesup:
3.1 vs. 2.04...
Hmm, where to start?  There are a lot of changes.  There's a new filesystem
(dircache).  Multiview and the datatype library and classes.  A bunch of WB
disk enhancements (to mount, the layout of things, etc).  HDtoolbox is
new-look.  Of course you get all the 2.1 stuff too (locale in particular).
Check magazine articles for a more exhaustive list.  3.1 is a Good Thing.

MarkM/MOD:
Availability?  Can you comment on that?

Randell Jesup:
Availability... That's really a marketing thing, including how it will be
made available.  It should be soon, though.  It's quite stable.

jon:
What will the graphic and animation capabilities of the Next Generation
Machine be like? Will RTG/DIG be a part of it?

Dave Haynie:
So, ya wanna talk graphics and animation.  I could write a book on this.
In fact, some have.  In general, bigger, better, faster, more.  You will
have an improved (eg, faster) blitter.  Graphics hardware handles chunky
pixels, 16 and 24-bit direct color.  Non-interlaced resolutions can go up
to 1280x1024 (not necessarily at 24-bit, however).  RTG will be an integral
feature of the next generation, and in fact necessary to handle chunky mode
pixels, for instance.

jon:
Will RTG be homegrown, or could EGS be adopted?  Its here.

Randell Jesup:
RTG: We haven't ruled out EGS - I'm not the primary gfx guy, so I don't know
all the details.  However, most likely we'd want something that kept as much
as possible compatibility with current software and calls.  I 'm not sure if
it gives the most for that, because I (personally) haven't looked.  The GFX
guys are doing that.

Fred Murray:
The latest setpatch (40.14 I believe) had a fix for A600s with Conner hd's.
Is there any chance a patch can fix this slow Seagate (st-914) slow seagate
in my 1200?

Randell Jesup:
We can't easily make a slow drive faster, unless you have a good trick for
making time run faster.... ;-)

Fred Murray:
Well, 200k/sec reads are quite slow!

Randell Jesup:
If there's a specific problem, perhaps.  However it would hav e to be a
major
problem.  Note that many 2.5 drives _are_slow.  They weren't designed for
speed. If you need a fast 2.5" drive, buy one to start, or use something
like
external caching software.

Paul Idol:
Will the 4000, or the rumoured low-cost 4000, be upgradable in any way to
future architectures and OS versions, like AAA and OS 4.0, or whatever
comes after 3.1?  For example, will Zorro II and III cards be usable by
future machines and vice versa?

Dave Haynie: Yes.

Paul Idol: How completely?

Randell Jesup:
New OS versions are usually runnable on older machines.  That will continue
so Zorro is Zorro, as far as we are concerned.  As long as it's possible.

Paul Idol:
Well, I heard about PCI - and 3.0 was never released for AGA machines.
And what about AAA for 4000s?

Dave Haynie:
Future machines with Zorro slots will run faster Zorro III, but that's a
controller/system interface function.  It does not impact on a card's
design.
I am currently in the process of looking into adapting some evolving
technologies for A4000 use, officially.  Like I tried to point out earlier,
a major new system doesn't happen all at once.  Since our next generation
architecture is modular, pieces can be adapted for A4000 use before the
A5000 is ready.  I expect this will happen.  PCI is a long term key to low
cost modularity.  Back in 1991 I started working on the post-A4000
architecture.  I designed a "modular interconnect bus", which I  called the
AMI bus, for this purpose.  Later in '92, PCI was unveiled, and it no longer
made any sense to go the custom route.  Still, PCI or AMI, the main point of
this design is to support on-motherboard modules, like graphics, CPU, etc.
It's a local bus replacement.  It does allow us to make an intermediate
machine expandable via a PCI slot or two, but that's about the limit on free
PCI slots.  I expect a full blown slotted Amiga would also have Zorro slots,
much like PCI-based Clones have EISA or ISA.

Steve Ahlstrom:
Dave, even tho you are working on new hardware, do you see any signs that
CBM is interested in anything other than CD32?  Do you have the manpower
necessary to develop both hardware and software for future computer
products?

Randell, you say you aren't the primary gfx (software) guy... who is?

Dave Haynie:
Steve, I have needed more manpower ever since I started at C= back in '83.

Randell Jesup:
Allan Havemose is head of Amiga software, and is covering GFX until we hire
more GFX people. (He used to be head of the GFX group).  I'm in charge of
the OS group (ie. everything not GFX or UI, basically).

Steve Ahlstrom:
Ok .. guess I'm asking if there is an upbeat feeling or are you guys looking
for jobs?

Dave Haynie:
We have enough folks on the high-end, and a few we share with the low-end,
to do what we need to do.  I would like more, it would make the "A5000"
happen faster.  However, like I mentioned, you don't have to necessarily
wait for the A5000 to see the fruit of our next generation labors, if all
goes well.  You do have to have every piece in place to get an A5000,
obviously.

Randell Jesup:
We have posted on the Internet requests for resume for GFX people, and may
well be posting more positions soon (in software).  In GFX, we also have Ken
Dyke and Fredrick Shaw  (a new hire from Ensoniq).  Obviously, things have
been better.  However, sales are looking up with the CD32 introduction and
1200 sales.

Stuart H. Brand:
Will there be a return of speech synthesis (localized) or perhaps voice
recognition, or are these better suited to 3rd party developers?

Randell Jesup:
Speech synth: we've been negotiating with some people, so you may see it
re-added (and better), as well as possibly non-English languages.  Many of
our machines are sold to non-english-speakers, and the old narrator didn't
help them much.  However, no promises.  We don't have a lot of money to
throw around, and I don't know if it will happen.  As for recognition: we'll
leave that to third parties.  It probably requires a DSP or a very fast
processor.

Mike Smith:
What part of A5000 will we see first and when?

Dave Haynie:
The main interests seem to be adapting "A5000" CPU and graphics subsystems
to the A4000.  The next generation CPU subsystem is perhaps the simplest
adaptation, though we're technically further along in graphics.  I can't
really predict which will get out first.  Also, as I mentioned, our DSP
technology has been reasonably solid for a year.  It has been licensed out
to third parties, and if we do decide that's the best route  for motherboard
audio in the it may wind up adapted to the A4000.  In engineering, I'm
responsible for telling the company what's possible and "making it so" when
they have decided on the course of action.  Of course, I do attempt to
influence the directions the way I see fit, but I don't have total control.
Once a complete "technology" is done, adapting it for use on a card may
be accomplished in a matter of months, so it's not like stuff that is
now working will have to wait 'til '95 or anything to be released, if that's
the course C= decides to steer.

Michael:
Forget about everything you _really_ know about CBM, and forget about who
you
work for.  Would each of you tell us what three things YOU would MOST like
to
see in the NG machine?

Dave Haynie:
Ok, me first.

[1] Graphics.  I have an oMniBus card here on my Amiga.  It does
1180x900 noninterlaced, but its not fast.  I want to have state of the
art graphics on the next generation system, coupled with RTG.
"State of the art" is sometimes a matter of months, that's where RTG
comes in.

[2] Modularity.  Since the A2000, I have been a fan of modular
systems.  Look at the A2000.  By the time it was done, you had a
tricked out A2500 that ended about where the A3000 began.  I believe
that's the way it should be done, and I hope I can return to this
philosophy w.r.t. the A3000/A4000 vs The New Thing.  Also, I think
new architecture can make modularity nearly free -- the cost of
modularity has also been a concern.

[3] DSP.  I worked on the DSP project for about a year and a half, and I
seriously believe it is the way sound will be done in the future.
I think even the conservative PClone industry is thinking this way.
Good sound effects in a program are probably done just as well with lots
of DMA channels.  The real advantage of the DSP is the sounds I expect to
hear once some good audio hackers have 25-33MFLOPs to play around with.

Randell Jesup:
1) Graphics.  One guy here has a Picasso 2.  I want it.  Resolution
is _wonderful_.  I use a Moniterm, and find 640 or 800 too narrow. As Dave
said, to be able to get state-of-the-art you need to have good RTG.

2) CPU.  As with most software people, I want _speed_.  More
CPU means we can build software faster (since we can layer it more with
out performance problems), and more generically (classes, things like data
types, etc).  Obviously we can get faster Motorola CPU's (faster '040's,
 faster memory systems, and the '060 is coming soon from them.
In the long run, RISC is the only way to go.  I've been promoting RISC
Amigas around here since '89 or '90.  Back at GE Corporate Research, I
was a member of a RISC CPU development team (the RPM-40).

3) hmmmm.
I usually don't think about it this way.
I'd very much like the see the DSP stuff come out; I was the software
"contact" for it (Eric Lavitsky was doing most of the sw work under
contract).  I'd also like to see improvements in the OS.  Unfortunately,
it's become hard to make fundamental changes to the OS due to compatibility.
For example, protection is a major pain to try to do under Amigados.
VM is possible, if a bit kludy.  Eventually we may have to make some level
of break with 100% compatibility to move forward.  However, don't expect
that
too soon, and it may be tied to CPU issues.  Oh well, I've rambled enough.

Michael:
Now, on a scale of 1 to 10, how much credence does CBM give _your_ ideas of
what's important, as you just outlined?

Dave Haynie:
Letsee here...

Randell Jesup:
Both of us meet with Lew fairly often.  Our ideas are fairly in accordance
with his, I think.  That doesn't mean we'll get all we want, necessarily.
I think these broad sweeps we've given are both pretty obvious and likely.

Dave Haynie:
[1] I think Lew sees eye-to-eye with me on many of these issues. The overall
next generation plan I worked on starting back in '91 or so was picked up on
by Lew and some advanced research stuff he was working on with Ed Hepler
(our main Chip group advanced architecture guy) independently of what I was
working on.  Yet, when presented in some decently complete form, both ideas
were not only along the same lines, but had the hardware magically appeared
before us, they would have played together.

[2] see 1.  Graphics is the priority, plain and simple.  That's with good
multimedia support, we want to do things the Amiga way.

[3] The issues aren't settled by any means, but things do seem to be warming
to DSP.  I think a realization of our DSP technology in the market by 3rd
parties in the immediate future, plus standardization on DSP as the next
generation sound device in the industry (Apple has already, but it'll take
the Clone biz to seal the deal) should promote my goals.  After all, it
already works, and the simple things like "play a sample" or "record to
disk"
are essentially built-ins under VCOS/VCAS.

MarkM/MOD:
Let me ask one final question... you have heard the doomsayers.  Many Amiga
owners are depressed.  I think this CO will help a lot by the way. Is there
anything you would like to say to the Amiga community in general?

Dave Haynie:
Sure.  I'd just like to wax philosophic for a second or two.
I think the only way to address a problem of any kind is to move directly
into it.  Perhaps its too much Aikido practice, but I think any other way
you're doomed.  That isn't necessarily C='s business practice, but it's in
my approach to Commodore.  I look at the technical problems, business
problems, etc. and make a decision.  Do I attempt to address the technical
problems or do I forget about Commodore.  At this point, I choose to address
the continuance of the Amiga, and I do so because I believe that it is
something worth doing, and something that will yield success. If I did not
believe this, I would have left the company, plain and simple.  Obviously,
not everything is under my control, but I know what I have to work with over
the next year or so, and I judge it adequate to achieve the necessary goals.

Randell Jesup:
Personally, I don't listen to the doomsayers... ;-)
Commodore has gone through a rough stretch recently.  However, things
are starting to look brighter.  CD32 has done pretty well in Europe for
a brand-new machine (far better than 3DO has done here).  A1200 sales
have picked up.  Financially, Commodore is in better shape than it was.
Not good shape, but better.  The cost-cutting, while very painful, has
given Commodore the time it needed to launch the CD32, and to continue
work on next-generation products.

I've been here since '88.  Of people in software, only one (Eric Cotton,
who manages tools and releases) has been here longer (he dates back to the
VIC-20 days).  People may not remember, but the software group was much
smaller than it is now when I got here (4 people).  Also, a number of
senior people who've left recently have gone to places like Scala, so
I wouldn't say that's entirely negative.

Also, if you follow the cable-box stuff in EETimes, etc, you'll know that
we're a player in that whole thing, since we bring a lite OS and good
NTSC graphics.  Think of an Amiga in your information-highway set-top
terminal... ;-)  There may be clouds, but a lot of them are behind us.
Certainly there are more hurdles ahead, but I think we can handle them.

*******************************************************
*
* Was there more?  Yep.. sure was!  Another 3 hours or so!
* Those 3 hours were not part of the formal conference so they were
* not recorded for editing.  Dave talked about the differences
* between SETCPU and CPU.  He talked about the future.  The only way
* to see the whole story is to join CompuServe and be there when it
* happens!
*
* This file may be freely distributed in any medium so long as the file
* remains intact with no changes -- including this notice.
* The conference was moderated by Mark D. Manes.
*
* If you wish to subscribe to CompuServe simply dial 1-800-787-RUSH.
*
* Copyright 1994 AForums Ltd.
*
*******************************************************

------------------------- Cut here ---------------------------------


--
----------------------------------------------------------------------
                             | "Ah Hein ... Reginald you have the
 [e-mail] |  wrong map here you silly old
                             |  leg-before-wicket English person."
----------------------------------------------------------------------