From: dave.haynie@scala.com (Dave Haynie)

Subject: Re: The Amiga.. some questions, suggestions and gripes.

Date: Tue, 18 Apr 1995 21:31:43 GMT



In <19950415.1468A8.649A@gandalf.eastcoast.co.za>, misterx@gandalf.eastcoast.co.za (Myron Alexander) writes:



>With the demise of commodore, most of the brilliant engineers left for

>paying pasteurs,



Actually, all of the Commodore Engineers who hadn't left already were

kicked out last November. No engineers remain with the ghost of

Commodore.



>2) How good are the remaining hardware designers. I hope that we will

>   not have any more lack of foresight. An example of such lack of

>   foresight is with the 31kz mode capabilities of the AGA, why did no

>   one think of placing a switch that forced the 31khz mode.



Someone did -- me. I actually suggested we use one of the six jumpers

remaining in the Lisa "extended mouse port" register (the first two

are used to tell if the AA system is 16 or 32 bits wide, and whether

it handles burst or not). The reason this wasn't done was the lack of

ROM space; there's no room in the ROMs for a 31kHz driver. I wasn't

happy with that, but that's the story.



>3) If Amiga Inc. or CEI (don't know anything about them) buy commodore,

>   would some of the engineers that have left return to the amiga flock.



That's a good question. I think the new folks might manage to get a

few back if they hit the ground running, and can demonstrate they have

a comittment to the Amiga, and the money to back it up. I don't think

any of us want to go through the end of the Amiga all over again.



>4) In a recent article in Amiga Format, an amiga engineer was

>   interviewed about the developments of the AAA+. He said they had

>   scrapped the original AAA and were now working with H.P. to

>   integrate the AAA (or whatever) into a HP PA-RISC 150.



That's not quite the story. What they're talking about there is the

Hombre project. Hombre was a brand new chip set design, started maybe

two years ago. It is 100% incompatible with the Amiga chip set. It was

never built, but the design at least of the control chip was fairly

far along when I left C=. This chip integrates an enhanced PA-RISC

core, VRAM controller, new style blitter, etc. When coupled with the

display chip, it can act as the core of a high performance 3D game

machine, or a fast RTG graphics board. I/O in both directions is over

the PCI bus, and an external PA50 can be added to boost performance

for mid-range systems.



>   I managed to read a copy of Dave's original AAA specs

>   released privately at a show in 1993/4 (not sure) and what was

>   being proposed was absolutely breathtaking.



You probably saw my writeup of AAA for the 1993 DevCon in Orlando. I

didn't design any of AAA, but I did design the prototype machine it

plugged into.



>Dave, in your opinion (based on a heck of a lot more info that I

>would ever hold) just how good does this new AAA system compare with

>your original specs and will there be a seperate processor for

>program/OS execution.



Again, Hombre isn't AAA. Graphics-wise, Hombre is better than AAA in

most graphics things. It's faster (well, it does have a RISC chip in

it), supports 3D operations, better blits, etc. It was designed

primarily for 16 and 24 bit graphics, and supports four 16-bit

playfields at once. I think the top resolutions on Hombre were the

same for the 64-bit AAA configuration, 1280x1024.



But it's not Amiga compatible, and it's not a full system solution. At

least when I last saw the specs, it was a graphics system only; it

didn't address floppy, ports, sound, etc., although there was a

proposal to adapt some of the sound subsystem from the Mary chip (AAA

analog of Paula). That's not necessarily a bad thing, anyway; in a

game system, you need some kind of sound chip, but probably whatever's

closest to standard. On a new computer, I would expect a device

independent specification for hooking in digial sample I/O and General

MIDI I/O, nothing hardware-specific.



The SYSTEM architecture I was designing as a replacement for the A3000

architecture (which formed the basis of the A3000, A3000T, A3000+,

A4000, A4000T, and Nyx, the AAA prototype), was designed to handle the

graphics as a plug-in module. So it would work with AAA, it would work

with Hombre, or with any off-the-shelf PCI-based graphics card.



>9) As a potential developer, it would be nice if a subset of the autodocs

>   were released PD.



That would be nice. I guess the problem right now is that there's no

one at the former Commodore with the authority to get something like

this going, even if they had the inclination. Certainly all such

things, like copyrights, etc. are frozen until the liquidation. Of

course, there's nothing stopping anyone from writing a PD book which

includes the same level of information, just as there's nothing

stopping anyone from writing such a book commercially. You just need

to be familiar with the difference between "using reference materials"

and "plaugerism".



>10) Wouldn't it just be grand if the new amigas came network ready.



Yup. I had that in for the A3000+. It supported a 2MBit/second network

using RS-485 media and arcnet protocols. By the time we were into it,

though, a faster version of the network was out, which supported

4MBit/second rates. That would be effectively about as fast as

Ethernet, since Arcnet uses token passing rather then the collision

detection that tends to slow down an Ethernet setup.



>11) Do away with the A1200/600/500 style casings. They look cheap and

>    belie the power under the "hood".



Nope. One of Commodore's strengths was the low-end, and you're always

going to be able to deliver a lower cost machine in that kind of case.

Or, to put it another way, they should always have the lowest priced

machine in that kind of case until a cheaper alternative comes

along. Especially at the low end, what matters is what's inside. If

anyone cares about the look, they're welcome to buy a different

model.



I do support there being more models. We tried on several occasions to

deliver a mid-range Amiga, something squarely between the A500/A1200

class and the A3000/A4000 class. The last one that was cancelled was a

system called the A1000+. This was a AA system that was to follow up

the A3000+, but using its own cost efficient system architecture. The

idea was to deliver an $800 25MHz AA system, in a low-profile "near

pizza box" case.



>12) Following on from 11, get a decent designer to design the casings.

>    Apple, Microsoft and others have discovered the importance of

>    making a computer/peripheral look very appealing.



The right thing to do is use standard, everyday PC Clone cases. That's

the only way to be cost effective, except at the very low end. As an

individual, I can buy a PC Mini Tower, quantity 1, with 200W supply,

for about $35.00. That's less than half of what Commodore paid for the

200W A2000 supply, alone. That's the cost of making custom parts; if

your competition isn't doing it, you can't afford to either. The

exceptions are when you're making a million of something that's

inherently cheap to make in volume. Like a plastic A1200 case.



>Take a long hard look at the Macs,



Macs had fancy cases since the beginning. Then again, Apple used to

get about 5x markup from cost to list. You'll notice they still use

custom casework, but it's looking more like PC casework every day.



>the Silicon Graphics Indigo



Silicon Graphics sells a rather expensive box, compared to a $1000 or

so personal computer. They can afford a custom case, and they would

probably have a hard time using the PC casework. Then again, other

R4x00 and Alpha based computer makers use the generic casework.



>and the "Microsoft" Dvorak keyboard.



Microsoft sells some of the most expensive keyboards on the

planet. They can afford to make them look fancy.



>I personally love the design of the CDTV.



Yeah, it looked nice. But it cost so much to make, no one bought

it. That's a luxery few companies can afford. Commodore certainly

couldn't.



>It was the first time a Amiga looked professional.



Professional? The CDTV looked like a piece of consumer stereo

equipment; there's nothing professional about that.



>14) To software developers. Why is our software sub-standard.



What's the standard? Some Amiga programs, like spreadsheets and

wordprocessors, probably don't quite match up to their PC

analogs. That's probably because the world's largest software

companies write the PC versions, while small companies write the Amiga

versions. I'll bet the 4 or 5 people on some of the big Amiga programs

are better programmers than the 25 or so people on a big Microsoft

program, but it's nearly impossible to do the work of 5 people, even

if they're average ones. Believe me, I've tried.



>    On the PC, apart from video, I can't fault their wordprocessing,

>    or spreadsheet, or DTP, or database,



I guess as long as "I don't mind the bugs" is your motto, Microsoft

stuff is not to be faulted. At Scala, we have to use Word, the company

adopted it as the standard. It's very powerful, and even a moron could

use it. However, it's also the buggiest program I have ever used on a

regular basis (I have seen a very few worse, but I refused to use

them).



On the other hand, there are other pieces to choose from. It's simply

the size of the marketplace. There are what, 5 million Amigas, most of

them low-end. There are around 40 million new PCs sold every year,

and they don't always replace old ones. I have no idea what the total

is, I'm sure the estimate is up at a couple 100 million at least.



>    Also, I seem to be getting more speed out of Wordperfect 6 on a 386

>    DX-40 than Final-Writer on a souped up A1200 with '030 at

>    50mhz. And the 386  is being bogged down by Windoze.



How much fast RAM is on that '050, and just how fast is it? How much

level 2 cache? In a fair race, an '030 will edge out a '386. But I

doubt you have a fair race set up here. There's a good chance the '386

has a faster graphics card, too, which makes all too much difference

in the speed of a wordprocessor.



>    Lets not talk about the speed of Pagestream or the so called

>    graphical superiority of amiga and its software.



Don't compare DTP's to WP's, either, they're two different

animals. Compare PageStream to Pagemaker, Frame, etc.



>    Who would want to buy that commercial effort from HiSoft after

>    those [screen blankers].



Well, I'll tell you, I was kind of amazed that anyone, anywhere, paid

dime one for a screen blanker. Yet, the PC crowd loves them, and I

suspect a few hackers became millionaires with them. You can't fault

HiSoft for trying; it's a slam-dunk, and maybe not everyone knows that

much about what's freely redistributable. Or has the time to hunt

around.



>P.S. I heard a rumour that GVP is closing down because the owner has

>made enough money and wants to retire. Is this true ??



It's more like, "they didn't adapt to the changing market, and now

there's no money left".



Dave Haynie          | ex-Commodore Engineering |  See my first film

Sr. Systems Engineer |      Class of '94        | "The Deathbed Vigil"

Scala Inc., US R&D   | C= Failure n. See: Greed |     info@iam.com



 "Caught a bolt of lightning, cursed the day he let it go" -Pearl Jam