The Open Amiga, the Amiga Industry Council and Other
Ramblings
by Dave Haynie
Shortly after the fall of Commodore, I think it was June of 1994, I
went to a few Amiga shows as a guest speaker. Being nice guys all,
they had asked me a few questions in advance, which gave me something
to talk about. And given the then-dire straits of the Amiga industry
(if only we knew), there was quite a bit of concern expressed in these
questions. Commodore screwed the pooch, so to speak, and that was
especially scarey to those who figured Commodore was murdered by the
PC (of course, it was a suicide). So in answer to these questions, and
some of my own, I pondered the state of the Amiga, what went wrong,
why it went wrong, and what could be done to prevent it from ever
going wrong again. The idea of an open Amiga, and the Industry
council, were the result of this.
Keep Your Friends Close And Your Enemies Closer
Commodore failed, the Amiga's in decline, and the PC thrives. Why
this is the case is the topic of several books, not this thin
article. But the why of this is a critical question, and the answer to
that question a big part of some solution.
On the surface, and certainly a big part of the puzzle, is
Microsoft. They are an amazingly large company in the PC business,
they can single-handedly make or break an operating system or
computing platform, and their power is actually way beyond their
size. In retrospect, they got there through some good ideas and some
blind luck. The IBM PC was the first of that blind luck; IBM did about
as lousy a job on the PC as Microsoft did on MS-DOS -- a match made in
hell. But no one made computer all that great back then, and they were
often used for just one job. IBM's bad design wasn't inherently worse
than most CP/M machine or the Apple II, since it came later and could
benefit from new technology. Now, at 4.88MHz, the 8088 was
occasionally slower than Z-80 machines, which ran more optimized code
up to 6-8MHz. But the 8088 had two advantages. The first was memory,
the second was that empty socket; for $250 or so, you could get the
8087 chip. Enter Lotus 1-2-3, and an IBM PC could crank through
spreadsheets that VisiCalc on the Apple II couldn't dream of, 100's of
times faster. That launched the IBM PC, and Microsoft.
Microsoft's big break was actually based on how bad the IBM PC was
(what to want for six months of the best average big-company
engineering money can buy -- IBM was used to the idea that it just
didn't matter, they were IBM, after all). Thing was, as we all know,
anyone could make an IBM PC, it just look a few folks, like the boys
at Compaq, to figure this out, and soon the floodgates were open. MS
hitched themselves to this star by selling MS-DOS dirt cheap to you,
the manufacturer, IF you put it in every machine you sold. Basically,
they took a hit on short term profits, using that missing money to buy
themselves dominance in the market.
That's the story, really. There were two critical elements set up
here.
First, anyone can make the hardware. Once you have everyone making
the hardware, magic things happen. I'm at no risk buying a computer
from Compaq or Dell; if they die tomorrow, I'll get the same basic
thing from some other guy. Contrast that with Commodore, Apple, Atari,
or any other original system maker. The proprietary system hinges on
the health of one company. Amiga fans not only know this now, they've
lived it. So rule number one for a healthy Amiga market of the future:
Everyone can make Amiga compatible systems.
Then we take a look at Microsoft themselves. They thrived because,
unlike most of the early computer companies, they had Gates, who
turned out to be a smart marketing guy. Most computer companies were
started by techies, and it's quite possible you can't really
understand programming and marketing in the same human brain. Gates
was no software wiz, not even in the majors, but that wasn't
necessary. While everyone else was stupid, Microsoft wasn't
stupid. That was enough to make him rich. Occasionally they were
really smart, if perhaps in a common-sense way, and they planned
long-term. That, in this market dominated by technology companies that
couldn't give away free food to starving Ethopians, Gates thrived. I
heard he's currently worth 29 billion. That's like 4 million
times more than my bank account. So much for knowing the
technology without a stable marketing branch.
Which brings me to my second point. The industry needs a single OS,
and that OS needs to be driven by a company with some intelligence and
some thoughts for the next year and the next decade, not just the next
quarter.
To sum that up: There's one well handled OS company.
The Open Amiga
So everyone can build Amiga compatible systems. On the PC, this was
perhaps accidental -- nothing thrown together in a few months is going
to be all that difficult to copy. Or maybe they jus didn't care;
there was a longstanding tradition of copying IBM mainframe and
minicomputer hardware, so they just assumed it would happen on the
personal computer too.
Regardless, PCs could be copied. That architecture has been with us
in the computer business, like it or not, ever since then. It changes
gradually, thanks to a lack of central authority; it improves very
quickly, thanks to competition. It's ugly as the south end of a
northbound pig, but it's extremely functional. That doesn't mean it
can't be done better. There's a great fear among Amiga users that any
kind of improvement, any scheme that gets away from hard-wired
software-to-chip interfaces, will cause an inferior, un-Amiga-like
result. In a word: bullshit. Sure, it could be done wrong. But with a
proper set of interface defintions, no problem. Anyone who's used a
Cybergraphix or Picasso board in an Amiga knows it's still an Amiga,
even if the graphics engine was designed by some group at S3, rather
than Jay and the boys in Los Gatos many years ago. The Amiga's
characteristics are based on its software architecture, not a
particular way of building a graphics chip -- though the custom chips
were certainly the only way to get that kind of effect back in
1985.
An open Amiga will be defined by a hardware independent OS. What
this means isn't that there's no knowledge of hardware in the OS,
that's of course absurd. What it does mean is that every hardware
dependent bit is replaceable, and that the bits needed at boot time
are stored in a ROM on the motherboard, the rest being installed on
the hard disk. So as a hardware manufacturer, if I build the correct
ROM, provide the right device drivers, etc. I can boot up the vanilla
binary AmigaOS, at least for the processor I'm using, from a plain old
off-the-shelf CD.
There's a real concern, again and again, that things like RTG
graphics interfaces will kill the system performance. I spoke to this,
but since I know many of the faithful will eventually read this, let
me reiterate. This is not necessary, and very unlikely. One thing to
realize is that when we speak of "device independent", we're really
not talking about something like PostScript, where the output is so
independent of device that it'll look dandy on a screen, my 600dpi
DeskJet, or a professional's Linotype machine. What I'm talking about,
primarily, are software layers to isolate control functions.
If you know the programming, think about talking to a serial chip
on the Amiga. You open a device, make up a device message packet, and
DoIO() or SendIO() that thing. All of the control functions here are
abstract -- you have no idea if it's the Amiga serial chip, the A2232,
or something else. The data format is important, and something other
software figures out -- terminals, SLIP, or PPP, each use the same
serial device driver, and they go very close to the practical speed of
the system. Same goes with graphics -- control functions are isolated
with interface functions, data must be known to the program, at least
in some cases. If I want a 640x480 screen at 72Hz with 16-bit color
and chunky pixels, I should have a graphic function that opens that,
and either hands me a bitmap or fails. That bitmap could be
virtualized (eg, if I had 16-bit planar, it's possible to let the OS
translate it for me), or it could be real (eg, I have a framebuffer I
can talk to). Low-level primitives functions can draw a circle into
any bitmap type, using blitters if they're around. If it's rather, I
could bang the bitmaps, as long as program understands what's
available. But here's the low-level: control's abstracted, data
isn't.
Higher level software functions. Take the Amiga of today as a
queue; I can run PageStream in 1180x900 on my A3000, but I have to
fire up a normal Amiga-chip screen to run Scala MM300 or most
games. In a properly abstracted system, the games would write to the
low-level graphics, PageStream talks to Intuition functions, just as
before.
The New Owners
The new owners of the Amiga technology portfolio are, by
definition, the one OS company, until they decide to sell the OS, or
somehow subcontract its development. They have an amazing task in
front of them, and no matter what, they will fail if they try to go it
alone. Commodore couldn't keep the Amiga going as a 1 billion dollar
company, Apple's many times that size and having toubles because
they've been foot-dragging on the Open Mac. As I pointed out above,
the openness is a critical factor for survival.
The AmigaOS company needs to think about the AmigaOS and the
survival of the platform above all else. They also need to convince
hardware vendors that there's value in the AmigaOS. Maybe they can do
both hardware and software, as I presume they'll try, but that could
prove difficult, as it has for Apple, Be, and IBM recently. The
AmigaOS need upgrading, promotion, and porting, all in the light of
the fact that the new owners, most likely, spend most of their money
buying the Amiga technology. But that doesn't change things: the
AmigaOS will not survive, much less grow, on the basis of today's
technology, even if the fact of the new ower may tend to curb the
attrition rate somewhat. AmigaOS 3.1 and the 68K are not the answers
for most people in 1997. To survive, the platform needs new users who
stick around, and new developers. That will require ports to at least
one new technology CPU, it'll require OS advancements for user and
developer needs, and it'll require evalangizing like the world has
never seen.
The Industry Council
Because of the size of this problem, I don't think The New Owners
are going to be able to handle even the administration of the Amiga
Industry alone. And I've been there, done that, got a whole drawer
full of T-Shirts -- in a time when there were 100+ people in
Engineering, when CATS alone was about the size of PIOS today, there
simply was not enough time available for Commodore to do half of what
was necessary to maintain the AmigaOS.
Now, you may ask (so, g'head), "Dave, just what do you mean by
'support'". Well, glad ya asked. What I'm talking about here is
essentially architectural management of the AmigaOS -- what's the
standard for doing Thing X, what's the position on Thing Y, etc. The
New Owners may well be able to handle these, but usually, they don't
have the time to look into things they're not interested in. And when
they do, they may not have the time to apply true vision to their
solution.
Thus, an industry council. There's nothing quite like this in the
computer industry right now, but there are few pieces. For example, in
the general case of the industry, you have groups like IEEE or ISO, to
put together standards. Some go into the PC, some don't. The next
level up, you have VESA, which hammers out Video standards
specifically for the PC industry. Lately, after it's win against
Intel on NSP, Microsoft themselves have taken a lead on specifiying
the hardware as well as the software, in their yearly WinHEC (Windows
Hardware Engineering Conference).
What I'm talking about goes a bit further. I kind of envision a
sort of UN that helps coordinate the efforts of various different
companies. The New Owners are probably the only "superpower", based on
their ownership of the AmigaOS, they're naturally in the lead
position. Based on my assertion that the job is too big for any one
company, I would expect them to take a leadership position in "world
affairs". But they won't have infinite resources, and there will be
issues they can't immediately address.
One thing we have in the council is a way to create industry
standards. Any company might submit requests for technological
direction, or suggestions for technological direction. The council can
discuss this internally, but also distribute it to all member
companies as a council directive -- you don't know where it
originated. Companies working on similar things can work through the
council, again with anonimity if they want it, to show up what they're
doing. The goal here, of course, is to establish a common standard for
a thing, with as little repeat engineering as possible. Thus, it's
possible to maximize the efforts of individuals and companies across
the industry. This is a big missing piece in the PC industry, perhaps
necessarily so. Amiga companies will find that, while they're
certainly going to compete with one another, in reality it's the Amiga
platform against the PC. Anything that benefits the Amiga as a
platform ultimately helps them, anything that doesn't ultimately harms
them.
The council should be composed of a relatively small number of
people. It's likely there will be a permanent seat owned by the New
Owners, assuming they're actively participating. Other members can be
rotated through the other primary member companies, or via any other
means set up in the council's constution. While I think the primary
function of the council is a smart conduit between component companies
in the industry, that's kind of where it starts, not where it
ends. The council will meet some fixed number of times a year, and in
those meetings, future directions may be set. While it's unlikely
these will be binding on anyone, the member companies get the
recommendations through the normal channel, just as if these
technology requests or recommendations came from outside the
council.
The council should issue approved specifications and
recommendations for things, once it's come to that point. For example,
the council may decide for good reasons to recommend the CHRP
architecture, PowerPC processors, and Alpha processors. It would
present a position paper on each one, as an official document
available to any member, and perhaps those on the outside too. Other
companies could sign off on such documents with a letter of intended
support ("PIOS supports the PowerPC, and is building PowerPC systems",
"The New Owners support SCSI, and are including SCSI drivers for chips
X, Y, and Z in the standard OS release", etc).
Well, that's the basic idea. I don't claim to know all the
mechanics, but the idea is a fairly simple one, and it would be
useless to belabor the issue, or add complexity where it isn't
needed. As this grows into something real, I believe others will trim
the focus on the council's mission, and its relationship to The New
Owners and the rest of the industry.
The Logo
Another thing I thought up is an Amiga industry logo. If you look
around, you see these things on all kinds of equipment: "Compact DISC
Digital Audio", "PowerPC", "MPC2", etc. There should be a logo,
perhaps created by the council, that identifies a computer as
AmigaOS-ready, whatever that comes to mean.
Copyright © 1998 Giorgio Gomelsky