NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/programmer/6131

Internet Message Format | 1996-08-05 | 25.7 KB

Path: cwis.isu.edu!not-for-mail
From: [e-mail] (Skuld's Lover)
Newsgroups: comp.sys.amiga.programmer,comp.sys.amiga.games,alt.sys.amiga.demos,in,comp.sys.amiga.advocacy,comp.sys.amiga.hardware,comp.sys.amiga.misc,comp.sys.amiga.graphics
Subject: Re: AB3D II beats Quake....
Date: 24 Mar 1996 15:51:09 -0700
Organization: Idaho State University
Message-ID: <4j4jkt$[e-mail]>
References: <[e-mail]> <[e-mail]>
NNTP-Posting-Host: cwis.isu.edu

In article <[e-mail]>,
Fabio Bizzetti <[e-mail]> wrote:

[bunch of "PowerAmiga is an MS/DOS compatible!" stuff snipped]

>The *custom* (meaning lotsa times better) HP-PA RISC CPU was there!
>HP-PA and Commodore had a tight relationship, they already were working
>together since a lot of time. HP-PA has a future, and is perhaps the best
>RISC CPU available in a MultiMedia market. It is also one of the fastest
>FloatingPoint cruncher out there. The fastest HP-PA kicks ass of PPC620,
>that today is "not sure if we'll be available" because they discovered it's
>not practically faster than 604... that is not tha fast indeed.
>
>PowerPC has been designed by IBM: I dont need to add anything else.

Don't you understand why AT is working to make AmigaOS PORTABLE? It's so
you can USE AmigaOS on a PPC or a HP-PA or ARM or whatever you want! The
PPC port is coming first, of course, but AT have stated that other ports
will probably follow! PICK your CPU, it doesn't matter! Want to run
AmigaOS on your Pentium Pro? Sure!

[bunch of stuff clipped]

>Why??
>It's about time to make new custom chips!
>AGA is still 1983 technology, with x4 bandwidth (to RGB..).
>
>WHO SAID THAT IT IS NORMAL TO NOT HAVE NEW CUSTOM CHIPS????

Noone is saying that you CAN'T HAVE custom chips! Look at the
MS/DOS-compatible market! Many new models are SHIPPING WITH DISPLAY CARDS
WITH CUSTOM 3D AND GFX-PROCESSOR CHIPS!

The point of the PowerAmiga isn't to kill custom chips, it's to make it
INDEPENDENT of a single set of custom chips so we don't get stuck in the
same mire of hardware backward compatibility that saw us stuck with AGA
(and would've left us screwed with AAA, on which a LOT of old stuff
would've broken!)

The PowerAmiga will come with a bunch of PCI slots JUST WAITING TO BE
FILLED WITH A CARDFUL OF THE CUSTOM CHIPS OF YOUR CHOICE. All it needs is
a good game API to let you drive them without having to rape the
hardware. What? Virtual Ridge Rally USA doesn't run fast enough on your
stock EZ-3D card? Throw in the new VR-HedgeClipperMan-Bonzo-3D card and
watch it fly!

The hard part for AT will, of course, be to come up with a good API. One
of the features of a good API (and one that many of the megacorps' APIs
are lacking) is for it to be an OPEN API. One where the programmer, if he
finds that certain useful functions aren't implemented in the current
API, can add extra functions to it (and document the functions so that
others can add it to drivers for their own particular flavor of
hardware). The new functions could even be submitted to the authorities
(in this case AT) for eventual inclusion into the 'official' API.

A primitive but significant example of what I'm talking about is Gloom
Deluxe, or more specifically, the user C2P driver for Gloom Deluxe. The
coder of the driver is PERFECTLY WELCOME to rape the hardware to his
heart's content, because even if his driver isn't compatible with other
hardware, the documented interface is STILL THERE for someone else to
write their OWN driver.

An AT-sanctioned game API would include all the useful functions the AT
programmers could think of. After its release, they would examine and
adapt the home-brew modules that other programmers handed in, so the API
could continue to grow to suit the needs of more software.

>Commodore didn't die because they spent all their money to make new custom
>chips: Commodore died because they didn't spend money to make new custom
>chips.
>
>Is it so difficult to understand?

Commodore died as a result of many different mistakes. Lack of coherent
R&D <-> Production was one of them.

But if that were the only problem, the A1000 would've stormed the market
in 1985.

>Wouldn't the Amiga be the most sold games machine (with best market) if we
>had AAA when it was time to have it (lotsa years ago)? Wouldn't the Amiga
>(both high-end and low-end) have *no competition* if 3 years ago (when
>PowerMac did born), or also today, it had HP-PA Hombre?

No, not necessarily. Remember when the CDTV was released? It was the most
advanced console available at the time. It DIDN'T sell like hotcakes.

Commodore lost even when technology was on its side. The problems were
largely market related, unfortunately.

>Commodore died because they delayed the technology improvements.
>Commodore died because while the market (not only computers) had a huge demand
>of highperformances/lowcost chips, it missed all the opportunities.

Commodore died because the Amiga was never successful in dominating the
majority of the computer market, weakening C= so that what would be
serious but survivable fuckups for Micro$oft instead became fatal wounds.

>Watch "The Deathbed Vigil" about all the missed occasions.

Yeah. Does make ya mad, doesn't it?

But then, Dave Haynie was the designer of most of those projects that were
shelved. As I recall, you posted a message saying that Dave wasn't the
'visionary' everyone says he is.

>Escom/AT is sincerely acting *worse* than the Commodore of 1991.

But distinctly better than the Commodore of 1994.

>They're blind, accustomed to sell PC's to people that would buy *anything*.
>But the Amiga is not attractive for these standard users, and will always
>be seen as the most silly way to risk money and time.

I think you're getting AT and Escom confused here. AT operates ENTIRELY
INDEPENDENTLY OF ESCOM. Escom hands over the money, but they don't give
the orders.

>>Heh, are you implying that the OS isn't lacking?  How about these: >
>>* Virtual Memory.
>>* Memory Protection.
>>* Resource Tracking.
>>* A decent file system.
>
>Now they've to convert the whole OS from 680x0 to PowerPC, and it'll take
>ages, will be full of bugs: and the final result will not be considerably
>better than before, but much more complex.

So? We've got to cure our hardware-dependence at some point. It's a
painful step, but one that must be taken. Fortunately we're not the
first, and we can draw on Apple's experience to make our transition less
buggy and painful.

>Instead, adding the HP-PA subsystem would have made all *simple*, logical,
>immediate, simple, performant...

Noone is stopping you from dropping in an HP-PA on a PCI card. Design the
card, plug it in, and write the drivers.

You could even develop Hombre-esque functions for it.

>>* An attractive, consistent GUI.
>
>Perfectly handled by the HombreHP-PA, parallely.

With a good RTG API, perfectly handled by whatever capable hardware you
want, parallely.

>>* A good API and strong developers tools.
>
>Adding the HombreHP-PA would have made it practically 0 overhead, thus used
>by all Amiga games programmers: when a needed function wouldn't have been
>found in the API, the programmer would have created it, and released a
>patch for all users (and games programmers) that then would have been
>integrated in the next OS revision, rom or disk.

WHOAH! YOU UNDERSTAND! Sort of.

At first glance, this would appear to be exactly what I as saying above,
but aparrently you think that you can't have an API without Hombre.
That's weird, as the whole point of an API is that you can run it on ANY
hardware, not just a single custom chip(set).

A device-independent API doesn't NEED Hombre to develop. Hombre might be
a nice piece of hardware to write a driver for, but it would be cheaper and
easier to use off the shelf hardware (say a SVGA display, a normal [not
Hombre] HP-PA as a copper, and maybe another for 3D) to make a kick ass
copro/display card. But because we're doing this through an API, we don't
NEED that specific card, and the software would run on any other system with
varying levels of performance depending on the supporting hardware.

>Upgrading/changing the whole RISC subsystem would have been possible and easy.

Yeah! And you wouldn't NEED that SPECIFIC subsystem!

>>* Device independance.
>
>2 subsystems: Amiga 680x0 and RISC HP-PA Hombre (one or more), where the AGA
>would have been emulated by the Hombre (providing that it is not a fucking
>SVGA standard chip: it's the state of the art techonology, able to emulate
>AGA's copper as nothing).

Emulating AGA should be an option for users who are upgrading from older
Amigas. We shouldn't be FORCED into using hardware that doesn't suit are
needs (though a kick ass card fast enough to emulate AGA would certainly
be great for other things!)

>>Last but not least, where are our developers and publishing houses?
>
>As I said before, I (an extreme Amiga "fanatic") will not code for a *PC* with
>PowerPC cpu and no market rather than for a PC with a 80686 cpu and a huge
>market. Just because it has the sticker "Amiga" doesn't make it different
>than what it is: a *IBM PC*.

Then you must consider PowerMacs to be IBM PCs?

Try using a Pentium system running Windoze, then try a PowerMac. The
experience is totally different. Why? Becuase the OS is the soul of the
machine! It's what makes the machine behave in the way that it does. The
OS WHAT MAKES THAT PARTICULAR "COMPUTER" DIFFERENT.

You can think of coprocessors are essentially just software implemented in
hardware. Or you could think of support code as coprocessors implented in
software! Look at UAE, using it you can have an Amiga on a Pentium Pro!
Now admittedly the amount of resources wasted on implementing OCS in
hardware makes the overall experience not very impressive, but it shows
that coprocessors aren't "magic".

By standardizing how we interface with coprocessors, we can free
ourselves from being stuck to a certain line of coprocessors, even to the
point of handing the work over to the CPU if it has enough oomph to push
it.

The world is FULL of coprocessors, and while individually they may not be as
"cool" as AAA or Hombre, with a device-independent API we're free to mix and
match them until we can make a system that kicks the ass of any custom
chipset we ever dreamed of earlier.

The only way AT could ever HOPE to compete by becoming a custom chip
manufacturer would be to have a huge outlay of cash from Escom
(semiconductor plants cost a LOT, even to contract), and then produce
development tools for the chipset so that the resulting display card
could be used on any operating system. Of course then you'd see people
running Windoze on it and you'd yell and kick and scream that it wasn't
really an Amiga chipset then.

You see, big-name chip manufacturers like Cirrus make money by selling
chips to ALL markets, Intel-boxes, Macintoshes, DEC boxes, BeBoxes, and
yes, even the Amiga. They generate a whole dumpload of revenue that gets
poured back into R&D to make new and faster chipsets (they can't afford
stop and rest, as their competitors are developing too!).

Now imagine if Cirrus produced their latest kick-ass chipset ONLY FOR THE
AMIGA. Even if it destroyed everyone else, they'd soon fall behind
because the revenue from just the Amiga market wouldn't be able to fund
the kind of R&D that the other all-market companies would be wielding.

The reason the Amiga's custom chipset was so far ahead in 1985 (and AAA
would've been in 1992) was that the competetive graphics coprocessor
market is a very recent thing. Back then there wasn't any competition
because stupid marketroids didn't see a need for it, and small
dedicated groups of geniuses were able to design and release stuff that
blew away the competition.

In a way, small dedicated groups of geniuses could still design a kick
ass chipset on paper, but without a massive R&D budget to fund the
facilities for prototypes, and an even bigger budget to fund startup
production, and an even BIGGER amount of revenue from a large market to
allow the per-unit cost of the chipset to be competetive, there would be
no way for these small dedicated groups of geniuses to compete.

>Now imagine who never cared about the Amiga: what economical interests will the
>developer see in the PowerAmiga?? If he is mad, he can risk all for the BeBox:
>but also if he doesn't lose money and sells as much as practically possible, he
>will *never* earn more than he can do on the PC with Windows96 or anything else.
>And developing programs on Windows95 and then recompiling them on PowerAmiga
>OS, although not realistically possible without changes, will still be a crap
>conversion. A buyer gets a PC and gets the software instead of its shadow.

OK, but the same applies even if the Amiga had TOTALLY DIFFERENT
hardware? Why program for the Amiga when you could program for the bigger
market? Sure the Amiga might have a neat chipset, but there are neat
chipsets for other platforms too, and they would have device-independent
APIs so you could use the cool chipsets without having to worry about
compatibility.

People would write software for the Amiga because it offered a pleasing
environment, and that's what the OS provides. Take away the OS and you
have nothing more than a console. (Or even less than a console, if you
use the PSX, with its device-independent API as an example)

>Remove the Amiga's heart: you get nothing.

The heart of the Amiga is the lean, mean OS. Custom chips are icing on the
cake. They do lots of neat things but you can do the same things with
other hardware or hardware/software combinations.

>>> : The A1000/A500 and A2000 were dream computers IMO because of
>>> : their custom chips and architecture, not because of the lame
>>> : 1.0/1.1/1.2 OS.

(Oops, I deleted the attribute for these quotes... Oh well.)

Back in 1985/6 OS 1.0/1.1/1.2 was not lame. It was the most revolutionary
and amazing OS for the home computer market. By today's standards, OS 3.1
is more "lame" than OS 1.2 was by 1986 standards.

>>> Yeah, right... Maybe in 1985. Today the custom chips is the
>>> weakest part of the Amiga architecture.
>
>Again: why you talk about 1985 custom chips in the 1996?
>Who ever said that it's normal and sane to have them today?
>Amiga custom chipset+ is not SVGA, it's AAA, AAAA, Hombre. NOT SVGA.

GREAT. *YOU* figure out how we can make a new kick-ass custom chipset
with the resources AT has. And more importantly, *HOW* do we keep
competetive with newer updates while maintaining compatibility.

I'd be fucking ecstatic to see AAAA, or even AAA, but it ain't gonna
happen because there's no realistic way for AT to produce it without it
costing twice as much as the new 192-bit (I'm not kidding) ultra fast 3D gfx
chipsets that would blow "AAAA" away.

>It's sooooo naive to say that the Amiga has to become a IBM PC because the
>A500's gfx chip today are obsolete. Who ever promised "you'll have to use
>them for the eternity"? The OS evolves, the hard-disks evolve, the ram evolves,
>the cpu evolves. Now it's time to evolv the *custom* chips, not to throw
>everything out of the window (projects included) and buy a IBM PC, change the
>CPU, put the sticker "Amiga" and run a fake AmigaOS clone.

It's sooooo naive to think that it's possible to make these fancy new
chipsets with practically nil capital to start with, no production
facilities, no way of getting them, and finally expecting to pay off the
costs if you somehow managed to pull it off with only the revenue from a
single niche market of dedicated but few Amiga users.

It's also naive to think that hardware what makes a computer system.
Granted, hardware defines the maximum theoretical capabilities of a
system, but it's up to the OS to actually get the performance OUT of the
hardware into something we can use! Compare a Pentium running Windoze,
OS/2, and a UNIX clone (Linux, FreeBSD?). There's a world of difference
between them. They, in effect, make the machine a DIFFERENT COMPUTER.
They alter the capabilities of the system and the way we interface with
it. They can't give more than the hardware can supply, but they use the
hardware IN DIFFERENT WAYS to make the overall effect something
completely different!

>>I would actually say that at this point, the weakest part of the Amiga
>>platform is its manufacturer.  The OS and the hardware run an equal
>>second.
>
>>> What other _viable_ and _affordable_ CPU would YOU choose for
>>> the Amiga ?
>
>HP-PA due to the optimal relationship HP and Commodore had, and due to its
>extremely fast FloatingPoint capabilities, and due to its upgradability
>(HP-PA is a family of processors, all damn faster than most other RISCs),
>due to the unique possibility to customize also the CPU (as HP offers).

Great idea! When the AmigaOS is fully portable, you're welcome to USE the
HP-PA family. But that won't be for a while! Right now it's more
important to use the CPU which we can get up and running IN THE LEAST
AMOUNT OF TIME, and that's the PowerPC, because no other CPU has had the
tools for migrating from a 680x0 platform developed to the same extent.

HP-PA, DEC Alpha, MIPS, ARM etc. versions of AmigaOS come later.

>If you want more, then it's not of this world.

By the time the AmigaOS is fully portable, it may well be. ;->

>>HP-PA, Sparc, Alpha, in fact, name your processor... if only AT would cut
>>the AmigaOS losses, move a long way to portable source (instead of taking
>>the cheap-shot option with 68k emulation on the PPC).

The "cheap-shot" option is only a temporary crutch until the whole thing
is in portable code. Later on they'll be able to compile AmigaOS for
whatever CPU you want.

>Look: if you have a car, you can change motor and put a better one, and so on.
>AT/Escom waited when you weren't watching, changed the whole car, painted it
>again like the old one, and tryed to give you the same cockpit. This car is
>exactly the one you always hated, except for the cockpit and the color.

Who cares what kind of engine its running on? If the car goes faster than
the old one and costs less, I wouldn't.

Furthermore, the whole point of device-independence is to let you put
whatever engine you please in the car, and add whatever options you want
and have them all work properly.

And by letting you use standard cases, you can also pick your frame, AND
your paint job! :->

>I would complain hard.

I'd be happy at having a car with better performance without having to
lose the old interior that I'm so comfortable with.

>>> : Well,from the deepest of my heart: FUCK OFF AT,
>>> : you made it all become useless.
>>>
>>> Sure. This is just the constructive feedback AT needs. NOT!
>
>For what I've seen untill now, "constructive" and "AT/Escom" are two words I
>wouldn't put together without feeling a foolish.

They're a hell of a lot more constructive than Commodore turned out to be.

>"Have faith, wait more": this is old, and became obsolete now.
>There's a limit, when you see things how really are, there's no way anymore
>to hope.

Don't "have faith". Instead wake up and look at what parts of the Amiga
we can realistically save. And when I look at its obnoxious level of
dependence on expensive hardware that can only be produced by one company
that has no hope of ever surpassing the competition (unless it gets a big
influx of capital and sells to non-Amiga markets), I realize that we
can't save the Amiga-only custom hardware, at least on the hardware-raper
level.

What we CAN save is the awesome OS, which lets us get MORE out of our
hardware than the competition. All we have to do is get it running on the
same (likely) or better (less likely) hardware, and we'll wind up with a
better machine.

We can also use off-the-shelf chips in interesting combinations to make
neato custom chipsets, provided that we don't try to rape them and thus
lock ourselves into having to be compatible with them for all eternity.

>And my anger was completely justified by all the money I wasted and lost
>(instad of earning much more than the one I've spent) because I "waited and
>hoped".

Were you expecting to make lots of money on the Amiga? :-)

Maybe your hopes are being dashed by market realities, but the
alternative is to pursue a completely capitalist route and go off into
the sad world of Macro$loth, and I don't think you really want to do that.

Better to be forced to occasionally bow to market pressure than to prostrate
yourself completely at the feet of Emperor Gates.

>So mine was a very constructive remark: if AT is reading this, it should
>consider that many developers dont trust them anymore, because now it's
>clear what their future plans are, and only Escom's business likes them:
>they'll be the first company able to sell *IBM PC's* to ex "fanatic" Amiga
>users.

The day you take a Power Amiga, boot up the Amiga OS, and then wind up
seeing an MS/DOS prompt or a windoze logo, I'll believe you when you
call it an IBM PC. (Unless you're running PC-Task, of course! ;->)

>Sure some people will still prefer to hope to get the same Windows'95
>utilities compiled for the PowerAmiga, but this is all the PAmiga market
>and reason to exist. I believe Escom thinks:
>"He doesn't want to buy my PC? I sell him the one with the other CPU and
>the stickers that I bought nearly for free at that auction sale. WOW:
>I sell my PC's to every people, this is business".

You call a Power Amiga an MS/DOS-compatible with another CPU.

Then my A4000 is just an MS/DOS compatible with another CPU and a weird
A/V card soldered to the motherboard.

If I soldered AGA on a Pentium motherboard and ported the OS would it
suddenly be an Amiga?

>>Essentially, AT is giving the existing users that very message.  They're
>>alienating existing users in order to sell to supposed new markets.  Well,
>>here's where my unwavering support ends I'm afraid.  I'll keep my existing
>>system as my favourite old companion and novelty,
>
>I was waiting for PowerAmiga, but since it's a PC I just bought a 68060 for
>my Amiga, and will develop on my Amiga all the games for it and for 80x86 PC,
>to get some earning from this last (job) and to keep my passion for Amiga
>(hobby). It's much better to work on Amiga, developing and testing algorithms,
>making graphics and sounds, then re "compile" my HLA sources and data for PC.
>The 68060 is the best, I dont want a crap buggy slow PowerPC to do the same.

Er, don't call the PowerPC slow when even the slowest 601/66 is faster than
your 68060.

If you're really glued to AGA, you could just as well stick a PhaseV PPC
card in your A1200. :-) Whoah.

>>and I'm moving on to a
>>Sparcstation, and a portable Mac.  That way I can actually get things
>>done, and still play with my Amiga, writing a program here and there.  I
>>sincerely hope that AT finds a new market, and I regret that the direction
>>that they're following no longer suits my needs.
>
>Neither mine and of all the people that left the Amiga (I will publish all my
>games and programs also for 680x0 Amiga untill at least a single publisher
>wants them, (there're only 2 or 3 left to be sincere) because I will make them
>for my internal use though, also in the year 2002), and there's a lot of people
>that will leave the Amiga market soon just because it's always smaller and
>smaller

The shrinking market is unfortunate, but......

>while piracy is growing.

WHAT!?!?!!?!?! How the hell do you figure this?

I would think that all the hard-core pirates would go to a market where
there was more software to pirate!

Please explain yourself.

>While Motorola abandoned 680x0, Intel has not abandoned 80x86, that will never
>die IMO (I am technically sure about it).

Actually, pre-386 code will probably die with the P7 (it's already on the
back burner in the P6), and there's talk of making a complete shift to a
new architecture after the P7.

>It was time to *add* a HP-PA to the Amiga.

Once the code is fully portable, this will happen.

>IMO the PowerAmiga experiment will finish soon, Escom will reconvert the
>machines to run PowerMac software, still using Commodore name for PC's and
>Amiga name for RISC PC's.

I highly doubt they would sell an "Amiga" that didn't run AmigaOS.

> Names.

Operating Systems.

>The PowerAmiga will have more IBM chips inside than a Pentium has today.

Actually, the current run of AGA chips is manufactured at IBM plants
(Selectron and QuikPak do the PCBs and assembly)

>>The amount of support that AT is getting from the users which it primarily
>>intends to make money from is ridiculous, I've never seen anything like
>>it.  Most companies actually need to give something back to the users to
>>keep them loyal and ready to spend.
>
>>Peter Naylor
>
>
>PS: [nothing] Pet,Vic20,C64,Amiga,(philosophy changes),AGA-Amiga [nothing]
>
>Thanks Commodore, you've *been* the best of all untill business destroyed you.

The Amiga has nothing to do with the Pet, Vic20, and C64. Commodore
merely bought and altered the Amiga. They had nothing to do with the
basic architecture. The Amiga is NOT a descendent of the C64 aside from
the fact that a lot of old 64 users upgraded to it.

>BTW: It's official: Dave Haynie is *NOT* moving to Germany, nor working for AT.
>He personally corrected the rumor, saying an equivalent of "dont trust rumors,
>there's rarely something true in them".

Well, at least he's doing some consulting for them, so he'll have SOME
hand in the design of the PowerAmiga.

>|           Stop that fucking imperialist embargo against Cuba.              |
>|                 Hasta Siempre Comandante Che Guevara.                      |

BTW, I favor socialism as well, but we can't ignore capitalist rules when we
live in a capitalist economy.

Oh, and I can't wait to see Virtual Rally! ;->

-----
James Sellman -- Idaho State University           | "Lum, did you just see
--------------------------------------------------|  a hentai rabbit flying
[e-mail]   |    // A4000/40 14 megs  |  through the air?"
[e-mail]|  \X/  A500/20 3 megs    |    - Miyake Shinobu