An Amiga with the power of an sgi...
sl...@cc.usu.edu
been a lot of people talking about how they hope the AAA chipset will have a
chunky pixel mode. However, if Amiga development does continue by whoever buys
Commodore, there might be a better area to develop in.
As I've been watching the computer market over the years, I've seen
both the clones and the Mac, for the most part, come up to match the Amiga in
most aspects of graphics (colors, raw speed, etc). The Amiga does still have
the advantage of the blitter, programmable colors with the copper, hardware
sprites, etc. But these advantages aren't large enough to really get the Amiga
noticed. HOWEVER, even the low end sgi machines are noticed for, burned into
the very roms, are subroutines the programmer can use, much like those in the
Amiga's rom, but the sgi's built-in functions include texture mapping and many
other goodies, allowing for quick throwing-together of many sorts of graphical
projects.
Could you imagine buying an Amiga 5000 that had the AAA chips (16.8
million colors), 50 mhz 68040, and a 1 meg rom? In other words, the rom
expanded to twice its current size so that it held user-avaliable functions to
do texture mapping, 3 dimensional rotation, projection, and all other sorts of
graphics? The Amiga's strong point is supposed to be its graphics capabilites;
what if all a programmer had to do to add quality texturemapping to a project
would be to call a system function? It could possibly be as easy as supplying
a bit map holding the image to be textured, a viewport to be drawn to, and the
3d object to be textured to (set of coords or whatever)? Among BlitBitMap()
and FillArea(), add in Texture(MyBitMap, viewport, 3d_object);
The power of the sgi workstations is that they have standard the many
functions that are required for all sorts of graphics applications. I heard a
rumor (from a reliable source) that two programmers at sgi decided to try
writing a flight simulator with full texturemapping, etc. In one or two weeks
they wrote the flight simulator, using almost entirely system-standard
functions. Certainly an Amiga, while remaining less than a work station,
couldn't have as many standard graphics functions as an sgi, but just how much
could an affordable Amiga be stocked with?
Could you imagine just what a $2000 Amiga would be worth in the
industry that would have many of the graphical subroutines simulation and
graphics programmers need? With a computer as affordable as an A4000, those
programmers could quickly write systems off of all those functions they need
already existing, rather than having to write everything from bottom level.
How much would you (if you were a graphics programmer) pay for a
computer that had many (though not all) of the standard graphics subroutines to
produce all sorts of effects and simulations? I daresay, many would slit their
wrists to get ahold of such a machine without the sgi price tag.
Anyway, just my ramblings. If you want to flame me, go ahead =). Just
remember that I said an Amiga WITHOUT all the built-in graphic subroutines of
the sgi, but with many of them that are used the most. The wonderful people
who first desinged the Amiga stuck a whole lot on a 512k rom. If they expanded
it to 1 meg, just how much general graphics code could they shove on there?
-Jonathan
"That's BIG..." -Riker's amazingly intelligent comment upon noticing the Borg.
Michael D. Bayne
: Could you imagine buying an Amiga 5000 that had the AAA chips (16.8
: million colors), 50 mhz 68040, and a 1 meg rom? In other words, the rom
: expanded to twice its current size so that it held user-avaliable functions to
: do texture mapping, 3 dimensional rotation, projection, and all other sorts of
: graphics? The Amiga's strong point is supposed to be its graphics capabilites;
: what if all a programmer had to do to add quality texturemapping to a project
: would be to call a system function? It could possibly be as easy as supplying
: a bit map holding the image to be textured, a viewport to be drawn to, and the
: 3d object to be textured to (set of coords or whatever)? Among BlitBitMap()
: and FillArea(), add in Texture(MyBitMap, viewport, 3d_object);
You must note however, that the SGIs don't simply have blazingly
efficient routines for rendering polygons and such. They have
dedicated hardware for doing these things. No need to put algorithms
in ROM, one can just provide standard libraries. (In fact, AmigaOS 3.1
has an OS call ChunkyToPlanar() which uses the Akiko chip on CD32s
and uses a really slow software algorithm to do it otherwise). What
your argument drives at is that the Amiga needs better hardware. The
blitter and copper just don't cut it anymore. This is certainly the
case, but one has to consider the cost/performace tradeoff. The copper
and blitter have provided high performance at low cost.
I don't know if you've priced an SGI Extreme, but one of my profs
purchased one with 6 polygon rendering engines and it was upwards of
$30k. Granted, it's awfully speedy, but I'm not sure it would win a
rendering contest with $30k of networked Amigas.
--
+------+----------------------------+-----------------------+------+
| /o)\ | mba...@qualcomm.com | All that is is emacs. | /o)\ |
| \(o/ | #include <stddisclaimer.h> | Emacs is all that is. | \(o/ |
+------+----------------------------+-----------------------+------+
stephan schaem
: who first desinged the Amiga stuck a whole lot on a 512k rom. If they expanded
: it to 1 meg, just how much general graphics code could they shove on there?
What about a .library in libs: ?
Stephan
Zsolt Szabo
>I don't know if you've priced an SGI Extreme, but one of my profs
>purchased one with 6 polygon rendering engines and it was upwards of
>$30k. Granted, it's awfully speedy, but I'm not sure it would win a
>rendering contest with $30k of networked Amigas.
Was that an Extreme or Extreme^2? Anyway, the strength of an SGI is not
in rendering quickly alone; any "networked Amiga" could do that (as you
just pointed out).
On the contrary, I would like to see any Amiga do real-time raytracing at
1280x1024 resolution, in 24 bit (take a look at "Ideas in Motion" on even
the slowest SGI Indy and you'll see what I mean).
SGIs have been built for real-time graphics display; that's why they're
used professionally for CAD, etc. and not Amigas or any other computer.
ANYONE could write a program that does intricate rendering; doing it in
real-time is what the SGI is all about.
stephan schaem
: You must note however, that the SGIs don't simply have blazingly
: efficient routines for rendering polygons and such. They have
: dedicated hardware for doing these things. No need to put algorithms
: in ROM, one can just provide standard libraries. (In fact, AmigaOS 3.1
: has an OS call ChunkyToPlanar() which uses the Akiko chip on CD32s
: and uses a really slow software algorithm to do it otherwise). What
: your argument drives at is that the Amiga needs better hardware. The
: blitter and copper just don't cut it anymore. This is certainly the
: case, but one has to consider the cost/performace tradeoff. The copper
: and blitter have provided high performance at low cost.
Indigo/indy dont have specialized HW for 3d rendering. But still
kick compare to 3d GFX chip like the 34082.
I would be happy with an amiga corectly design around a 68060 (
meaning something not OCS++ based, or AA like some call it:)
: I don't know if you've priced an SGI Extreme, but one of my profs
: purchased one with 6 polygon rendering engines and it was upwards of
: $30k. Granted, it's awfully speedy, but I'm not sure it would win a
: rendering contest with $30k of networked Amigas.
SGI extrem can do realtime 3d, even if I give you 1million $ to
spend on amigas you wont be able to buy a setup to compete.
Stephan
Rainer Kowallik
In article <2vsv5a...@jhunix.hcf.jhu.edu>, robo...@jhunix.hcf.jhu.edu (Zsolt Szabo) writes:
|> In article <2vsfqd$i...@qualcomm.com> mba...@qualcomm.com (Michael D. Bayne) writes:
|>
|> On the contrary, I would like to see any Amiga do real-time raytracing at
|> 1280x1024 resolution, in 24 bit (take a look at "Ideas in Motion" on even
|> the slowest SGI Indy and you'll see what I mean).
I am not shure, that anything you could see there is realtime rendering !
If you meant the Introduction Demo Software, which starts when you
switch on the Indy for the first time (nice Idea by the way, could have
pushed the Amiga some years ago) it reminded me at some of the Demos
we have had on the Amiga (like the ancient juggler or some more modern demos).
They used only 256 colours and a low resolution screen ! (never thought, that
this was possible on a Workstation)
In fact, everything was taken from mpeg files
I didn't want to say, that the Indy isn't fast in graphics, but you
have to watch very carefully what is calculated in real time at what resolution
in how many colors. Otherwise you could take any Amiga Demo
and claim the Amiga (or the CD32 in mpeg !) as equally fast as the SGI.
Rainer
sl...@cc.usu.edu
> Michael D. Bayne (mba...@qualcomm.com) wrote:
> : You must note however, that the SGIs don't simply have blazingly
> : efficient routines for rendering polygons and such. They have
> : dedicated hardware for doing these things. No need to put algorithms
> : in ROM, one can just provide standard libraries. (In fact, AmigaOS 3.1
> : has an OS call ChunkyToPlanar() which uses the Akiko chip on CD32s
> : and uses a really slow software algorithm to do it otherwise). What
> : your argument drives at is that the Amiga needs better hardware. The
> : blitter and copper just don't cut it anymore. This is certainly the
> : case, but one has to consider the cost/performace tradeoff. The copper
> : and blitter have provided high performance at low cost.
>
so much any graphics coprocessor can do, and then the rest of the effects must
simply be produced by the computer's cpu.
When I wrote the original post, I didn't believe that the Amiga would
need specific coprocessors for graphics, but simply system commands (like
BltBitMap) that could be used for graphics and simulation programming. These
commands don't need to be ultra speed (have their own processors) they simply
need to work well, be generally usable for most graphics tasks, and do a good
job. Certainly, if someone was writing a game or a professionally-published
graphics package, they would write their own routines, but what about those
people to who speed wasn't the utmost important thing? They would have a lot
less coding at their hands to produce the same final product.
Jonathan Dinerstein
SL...@cc.usu.edu
Zsolt Szabo
:
No, as I said, take a look at the demo called "Ideas in Motion" or the
spider movie. Finally, try the SGI flight simulator which, at 1024x768
(or 1280x1024) does the smoothest animation I've seen.
Also, pay attention to the bust of Beethoven rendered and shaded in
real-time, transparently if must be. I think the framerate for that was
at around 30 fps, with 15000 polys/sec, fully shaded, etc.
Of course you could go all the way and check out the demos "Performer
Village", "Performer Town", "Distortion", "Atlantis", or even "Free
Flight"; all of these run even on a high-powered IRIS INDIGO (although I
should say that Performer Village/Town probably only run on the
Onyx/Reality Engine).
Can you say 1280x1024 resolution, RENDERED IN REAL-TIME(!!!) in 24 bit
color at 30 fps? In the case of Performer Town/Village what you see is
every DOOM-PLAYER's dream come true: a fully texture-mapped world with
everything to enhance the realism; even motion-blur is included and if
you get closer to textures they're blurred too (in real-time, of course,
because the blurring effect doesn't take place if you're at a distance)
so there is no pixelized look. Obviously the polygons are not orthogonal;
but this demo goes a step further than DOOM in that it creates whole
mountain ranges with trees/bush objects, etc. (all in 3d, of course).
Let me just say this: after seeing this I will never bee impressed by the
second release of DOOM!
Yes, the IRIS INDIGO I saw this on cost $26,000, and the Onyx, $200,000.
But I'll bet you that even if you bought $1,000,000 worth of Amiga
equipment, you could still not do this kind of animation with existing
Amiga video cards. The Amiga may be nice for genlocking and
frame-by-frame rendering, but for real-time display I would go
EXCLUSIVELY with an SGI. I've seen a SUN Sparc 10 ($20,000, if I'm not
mistaken) slow down to a crawl next to even the INDY ($8000), which was
designed to be cheap.
mis...@csc.canterbury.ac.nz
> need specific coprocessors for graphics, but simply system commands (like
> BltBitMap) that could be used for graphics and simulation programming. These
> commands don't need to be ultra speed (have their own processors) they simply
> need to work well, be generally usable for most graphics tasks, and do a good
> job. Certainly, if someone was writing a game or a professionally-published
> graphics package, they would write their own routines, but what about those
> people to who speed wasn't the utmost important thing? They would have a lot
> less coding at their hands to produce the same final product.>
of thing you are talking about, a lot of reasonable software functions to do a
lot of stuff, and when people need speed at the expense of generality they do
write their own. But what I wouldn't give for a blitter, the graphics chips in
some macs can do 200% scaling, but do you think its well documented, NOO, you
have to go through quicktime libraries to get at the functions to find out if
its there.
In my opinion machines should a combination of the two approaches, a
heap of simple to use routines and then some grunty hardware underlying it so
that when programmers really need speed they can get it.
Jon.
sl...@cc.usu.edu
> Sounds like you want a Mac. The toolbox on the Mac offers just the sort
> of thing you are talking about, a lot of reasonable software functions to do a
> lot of stuff, and when people need speed at the expense of generality they do
> write their own. But what I wouldn't give for a blitter, the graphics chips in
> some macs can do 200% scaling, but do you think its well documented, NOO, you
> have to go through quicktime libraries to get at the functions to find out if
> its there.
> In my opinion machines should a combination of the two approaches, a
> heap of simple to use routines and then some grunty hardware underlying it so
> that when programmers really need speed they can get it.
Uh, I wasn't quite meaning the stuff built into the Mac. What I mean
are more powerful stuff. Most programmers could throw together a scaling
routine in about two hours that would be faster than any system-standard one,
but it would take weeks to write exellent texture mapping code.
When I talked about routines the user could use, I meant advanced
stuff, subroutines that would take a deal of time to write. Thus, it would
be worth a possible slower speed of operation to a programmer to just use the
system command and not have to program it himself.
How often does an Amiga programmer use the command BltBitMap? Probably
a lot, even though owning the blitter and using it by hand would be faster in
specialized cases.
Jonathan Dinerstein
SL...@cc.usu.edu
Michael D. Bayne
Ok. Clearly, we can concur that the SGIs do cool animation and they do
it in real time. HOWEVER, you have to take into account that all these
demos of which you speak (and of which I've seen), are taking
advantage of a particular piece of expensive hardware in the SGIs.
I've looked at the GL library and they are simply setting up an array
of polygons to be rendered and setting a couple of parameters for the
Gouraud (sp?) shading model (have you noticed their nice little list
of surfaces, plastic, metallic, etc. are very blatantly Gouraud
rendering with specularity, ambient and diffuse components in
different combinations). Then they give their chips the address of
this array and the settings and BANG (or whatever sound a 150,000+
polygons/per/second chip makes when it runs) you have z-buffered,
shaded, blurred and texture mapped polygons and you have them fast.
That flight simulator didn't take long to write because it's just a
bunch of rendered polygons. They didn't have to do a whole heck of a
lot else. The demos that you speak of are like 3-liners. They're
REALLY simple because they do nothing but set up polygons and render
them with different parameters.
My point here is that the SGIs do a VERY specific task, Gouraud
shading of polygons, and they do it VERY fast. However, they do it
with VERY expensive hardware. You can't just whine and say, my Amiga
doesn't do that and won't for any amount of money because the Amiga is
not ABOUT costing $1 billion dollars to be able to render my butt in
real time. It has always been affordable. You are simply NOT going to
be able to afford to throw some nice polygon rendering chips into an
Amiga and get it to market for the 1500 or 2000 dollars that we've
come to expect. If someone wanted, they could very easily make a card
that you plug into your Amiga with some polygon rendering chips on it
and provide a library to use it. It's a simple DMA device. However,
unless capitalism has been wrong all these years, there simply isn't a
market for it yet. Certainly DON'T stick it in a stock Amiga and force
everybody to pay for some serious rendering power.
(Stepping off)
stephan schaem
: Ok. Clearly, we can concur that the SGIs do cool animation and they do
: it in real time. HOWEVER, you have to take into account that all these
: demos of which you speak (and of which I've seen), are taking
: advantage of a particular piece of expensive hardware in the SGIs.
The R3000 or rexx is not that expansive HW...(For example)
: I've looked at the GL library and they are simply setting up an array
: of polygons to be rendered and setting a couple of parameters for the
: Gouraud (sp?) shading model (have you noticed their nice little list
: of surfaces, plastic, metallic, etc. are very blatantly Gouraud
: rendering with specularity, ambient and diffuse components in
: different combinations). Then they give their chips the address of
: this array and the settings and BANG (or whatever sound a 150,000+
: polygons/per/second chip makes when it runs) you have z-buffered,
: shaded, blurred and texture mapped polygons and you have them fast.
: That flight simulator didn't take long to write because it's just a
: bunch of rendered polygons. They didn't have to do a whole heck of a
: lot else. The demos that you speak of are like 3-liners. They're
: REALLY simple because they do nothing but set up polygons and render
: them with different parameters.
Why do you need a machine that do 150,000 complex polygon second
to run GL or having an extended gfx library making sens?
And how otherwise should it be done? They give the address of the
'world/object' to their chip (s?), but its not a 3d chip its
to the mips(SGI) CPU. In the case of a basic indy/indigo the CPU
actually do it all but the final 2d rendering step.
(With Z buffer also handled by the cpu), for RE its an i860 array.
Like you just said GL is really powerfull, in '3 lines' you can
actually do great stuff.Thats why its kind of popular.
something like this should be avail on amiga, I dont beleive
it will be of any real use for A500 user, but still it would
be a great tool.And could become realtime on fast amigas.
: My point here is that the SGIs do a VERY specific task, Gouraud
: shading of polygons, and they do it VERY fast. However, they do it
: with VERY expensive hardware. You can't just whine and say, my Amiga
: doesn't do that and won't for any amount of money because the Amiga is
: not ABOUT costing $1 billion dollars to be able to render my butt in
: real time. It has always been affordable. You are simply NOT going to
: be able to afford to throw some nice polygon rendering chips into an
: Amiga and get it to market for the 1500 or 2000 dollars that we've
: come to expect. If someone wanted, they could very easily make a card
: that you plug into your Amiga with some polygon rendering chips on it
: and provide a library to use it. It's a simple DMA device. However,
: unless capitalism has been wrong all these years, there simply isn't a
: market for it yet. Certainly DON'T stick it in a stock Amiga and force
: everybody to pay for some serious rendering power.
GL do ALOT more then gauroud shading of polygon.
And on box like the Indy/Indigo there is no very
expensive specialized hw... actually there is no 3d hw.
I'm not saying the amiga should perform like a .2 million$ machine
but it could come around a basic indigo performance with a price
tag of under 2000$.
I personally think an 060 amiga could be priced at <2000$, and
beleive it or not... GL could run well just using the 060.
Maybe make the AAA blitter render 2d polygon with the possiblity
of mapping.Now this could give it another edge compare to PC
video card.(I cant beleive the 3do blitter is that expansive,
what about the jag GFX chip? , anyway)
Again the amiga DONT NEED a 3d chip, the indigo dont have one and
do pretty well.(When I checked, a basic 33mhz R3000 indigo rendered
faster with GL then a 40mhz 34082(3d math/gfx chip)+34020 with
tiga(Same concept you build a structure and voila its rendered)
AAA For bandwidth and chunky, 060 for rendering and you get
some correct 3d performance.
The idea is, make a 3d library avail for the amiga now(Speaking
like I was in the past here).So people use it, then they might want
more power and , just maybe all the 3d card that came out for the
amiga and died would have stayed alive has a rendering engine for
the '4d.library' amiga standart. Without SW HW is nothing.
Stephan
Val Kartchner
>(Stepping on soapbox, people with flimsy egos please look away)
>
>Ok. Clearly, we can concur that the SGIs do cool animation and they do
>it in real time. HOWEVER, you have to take into account that all these
>demos of which you speak (and of which I've seen), are taking
>advantage of a particular piece of expensive hardware in the SGIs.
>My point here is that the SGIs do a VERY specific task, Gouraud
>shading of polygons, and they do it VERY fast. However, they do it
>with VERY expensive hardware. You can't just whine and say, my Amiga
>doesn't do that and won't for any amount of money because the Amiga is
[It can't be done for the $1500 to $2000 that we've come to expect.]
[But someone could put it on a card for those who would like that.]
But the Amiga has been about powerful hardware for the price. When
the Amiga was introduced in 1985, it was doing higher resolutions and
a greater number of colors than other personal computers (I don't
mean just x86's). The Amigas closest competitor was the Atari ST.
What the Amiga was doing relied on *very* specific display hardware.
Here are some other things that the Amiga does (with OCS) that other
currently produced personal computers (of which I am aware) can't do:
* Calculate `life' with *very* little CPU intervention. The blitter
can calculate a new generation of `life' with 14 blits while doing
wraparound, and only 9 blits without doing wraparound. (Graphics
cards are increasingly getting blitters, but is it as accessible
from off the card as is the one built into the Amiga?)
* Change the palette on *every* scan line without CPU intervention.
(The Apple II GS can do this, but it's no longer in production.)
* Conceivably, the color palette and resolution could be changed in
mid-line without CPU intervention. (I've toyed with the idea of
creating a program which demonstrates this in a system-friendly
way.)
* Pattern fills can be done with little CPU intervention. (Another
blitter feature.)
I'm sure that there are others of which followups will remind us.
Nevertheless, that's what you can do by "taking advantage of a
particular piece of [in]expensive hardware in the [Amiga]."
In 1985, the Amiga was delivering superior graphics capabilities.
Through the neglect of Commodore management, the Amiga's significant
lead has been eroded. The AGA chipset now holds only a slight advantage
over the average SVGA card.
I don't expect the AAA chipset to deliver an SGI for U$S2000. However,
what should be achieved is hardware which will make the Amiga able to
provide a *much* better 3D environment in a system-friendly way at
that price point. Indead, the rumors for the AAA chipset point in
this direction. If true, this will carve out another niche market for
the Amiga which the new company will (hopefully) be able to exploit to
sell Amigas.
My point is that just because SGI's cost $20000+ to do what they do
doesn't mean that the Amiga can't be made to do impressive stuff in it's
price range.
--
|================= #include <stddisclaimer.h> ================/// KB7VBF/P11 =|
| "AMIGA: The computer for the creative mind" (tm) Commodore /// Weber State |
| "Macintosh: The computer for the rest of us"(tm) Apple \\\/// University |
|== "I think, therefore I AMiga" -- v...@csulx.weber.edu ==\///= Ogden UT USA =|
Ed Smith
sl...@cc.usu.edu
> Michael D. Bayne (mba...@qualcomm.com) wrote:
> : (Stepping on soapbox, people with flimsy egos please look away)
>
> : Ok. Clearly, we can concur that the SGIs do cool animation and they do
> : it in real time. HOWEVER, you have to take into account that all these
> : demos of which you speak (and of which I've seen), are taking
> : advantage of a particular piece of expensive hardware in the SGIs.
>
> The R3000 or rexx is not that expansive HW...(For example)
>
> to run GL or having an extended gfx library making sens?
>
> And how otherwise should it be done? They give the address of the
> 'world/object' to their chip (s?), but its not a 3d chip its
> to the mips(SGI) CPU. In the case of a basic indy/indigo the CPU
> actually do it all but the final 2d rendering step.
> (With Z buffer also handled by the cpu), for RE its an i860 array.
> Like you just said GL is really powerfull, in '3 lines' you can
> actually do great stuff.Thats why its kind of popular.
> something like this should be avail on amiga, I dont beleive
> it will be of any real use for A500 user, but still it would
> be a great tool.And could become realtime on fast amigas.
>
> but it could come around a basic indigo performance with a price
> tag of under 2000$.
> I personally think an 060 amiga could be priced at <2000$, and
> beleive it or not... GL could run well just using the 060.
> Maybe make the AAA blitter render 2d polygon with the possiblity
> of mapping.Now this could give it another edge compare to PC
> video card.(I cant beleive the 3do blitter is that expansive,
> what about the jag GFX chip? , anyway)
>
> Again the amiga DONT NEED a 3d chip, the indigo dont have one and
> do pretty well.(When I checked, a basic 33mhz R3000 indigo rendered
> faster with GL then a 40mhz 34082(3d math/gfx chip)+34020 with
> tiga(Same concept you build a structure and voila its rendered)
> AAA For bandwidth and chunky, 060 for rendering and you get
> some correct 3d performance.
>
> The idea is, make a 3d library avail for the amiga now(Speaking
> like I was in the past here).So people use it, then they might want
> more power and , just maybe all the 3d card that came out for the
> amiga and died would have stayed alive has a rendering engine for
> the '4d.library' amiga standart. Without SW HW is nothing.
>
> Stephan
What about a 3rd party company for the Amiga writing such a library?
Then those programmers out there who would want such could simply buy these
libraries to add to their compiler or assembler, and be on their way.
How long would it take and how hard would it be for three graphics
programmers to write general graphics routines? Couldn't three decent
programmers turn out a reasonable number of routines in a couple of months?
A library with a wide variety of graphics routines, priced at about $50
US, I would certainly buy. However, it definitely would be better if the
library were written by Commodore and incorporated directly into the machine.
Jonathan Dinerstein
SL...@cc.usu.edu
Andrew Gamlen
The SGI can't do real time ray tracing. It has custtom
hardware for doing polygon rendering in realtime and
the higher up models such as the Onyx can do real time
texture mapping. The Uni here just bought an Onyx
and some of the demos are unbelivable.
my 2 cents worth.
Zsolt Szabo
Let me ak you this: did your Onyx come with the Reality Engine add-on?
Ours did. Also, take a look at the "Ideas in Motion" demo. You will see
the SGI logo rendered, casting a shadow on the ground (with light
emanating from a fixed source, which has the shape of a lamp). Not only
does it produce shadows, even the surface of the logo is very plastic. In
other demos the computer calculates transparency in real-time as well,
with light sources of different color changing the color of the shadows,
etc. If you don't call this ray tracing, I don't know what is. Of course
it doesn't look anything like Lawnmower Man, but raytracing does not have
to be complex to be raytracing.
Judge ATomek
[... stuff deleted...]
> On the contrary, I would like to see any Amiga do real-time raytracing at
> 1280x1024 resolution, in 24 bit (take a look at "Ideas in Motion" on even
> the slowest SGI Indy and you'll see what I mean).
I don't mean to be picky about terminology but if there is a discussion in a
graphics group some things should be set right.
(1) there is no system in existance, that I know of, that is capable of
real time raytracing. "Ideas in Motion" doesn't use raytracing.
raytracing is not complex but it is extremely computationally expensive
"Ideas in Motion" looks really cool because SGI machines have some
really cool hardware and supporting libraries. It's FAST, again,
because of massive hardware support.
(2) term "rendering" is used to describe the process of graphically
representing data stored in your computer. It is not a method, like
raytracing is for example. Rather, raytracing is a method of rendering.
(3) FACT: SGI Reality Engine is capable of producing fully textured,
1280x1024, 30fps animations. This is the only system I have heard of
that is capable of this sort of performance.
cheers,
Tomek
P.S.
flames > /dev/nul
--
+------------------------------------------------------+
| /\ tm |
| /--\TOMEK tpi...@comp.vuw.ac.nz <-- New Zealand |
+------------------------------------------------------+
Dave Haynie
>> When I wrote the original post, I didn't believe that the Amiga would
>> need specific coprocessors for graphics, but simply system commands (like
>> BltBitMap) that could be used for graphics and simulation programming. These
>> commands don't need to be ultra speed (have their own processors) they simply
>> need to work well, be generally usable for most graphics tasks, and do a good
>> job. Certainly, if someone was writing a game or a professionally-published
>> graphics package, they would write their own routines, but what about those
>> people to who speed wasn't the utmost important thing? They would have a lot
>> less coding at their hands to produce the same final product.>
> Sounds like you want a Mac. The toolbox on the Mac offers just the sort
>of thing you are talking about, a lot of reasonable software functions to do a
>lot of stuff, and when people need speed at the expense of generality they do
>write their own.
No, the Amiga has what the Mac has in terms of a "graphics toolbox". It does
not have the same level of standardization for 3rd party graphics adaptors,
nor does it have 24-bit color support.
What's suggested here is the addition of something like SGI's 3D graphics
library. Strangely enough, some operating systems are adopting SGI's 3D
graphics libraries. Windows NT has, for instance. At present, most personal
computer operating systems support resonable 2D graphics, but the 3D
question is just starting to be an OS issue at this level. It's actually
more of an issue at present in game machine OSs.
In a related issue, if anyone's serious interested in getting SGI Indy class
performance on their Amiga, it's easy. Just run version 5.1 of SGI's OS on
your Indy...
Dave Haynie | ex-Commodore Engineering | Ki No Kawa Ryu Aikido
Sr. Systems Engineer | Class of '94 | "Life was never
Scala Inc., US R&D | "See us in the movie!" | meant to be painless"
Jamie Krutz
> Let me ak you this: did your Onyx come with the Reality Engine add-on?
> Ours did. Also, take a look at the "Ideas in Motion" demo. You will see
> the SGI logo rendered, casting a shadow on the ground (with light
> emanating from a fixed source, which has the shape of a lamp). Not only
> does it produce shadows, even the surface of the logo is very plastic. In
> other demos the computer calculates transparency in real-time as well,
> with light sources of different color changing the color of the shadows,
> etc. If you don't call this ray tracing, I don't know what is.
There you go. ;) It's not and you don't.
While you can use ray tracing to accomplish these effects, you
can also accomplish them in other ways. Most of the animations
in TV commercials and in films are not ray traced. They still
have shadows, reflections, specularity, etc.
-J
Zsolt Szabo
>While you can use ray tracing to accomplish these effects, you
>can also accomplish them in other ways. Most of the animations
>in TV commercials and in films are not ray traced. They still
>have shadows, reflections, specularity, etc.
Well, then would you please be so kind as to explain exactly what the
definition of raytracing is. For one, how do you suggest that transparent
surfaces can be rendered or objects can be mirrored in other objects without
raytracing. Because many SGI demos do both, sometimes at the same time.
Mark Thompson
|> Well, then would you please be so kind as to explain exactly what the
|> definition of raytracing is. For one, how do you suggest that transparent
|> surfaces can be rendered or objects can be mirrored in other objects
|> without raytracing. Because many SGI demos do both
Raytracing is a specific class of rendering algorithms in which beam paths
are traced between the eye point and the lightsource, with all the recursive
bounces those beams take along the way. It is the most brute force, slow, and
simple method of rendering a scene. SGI nor hardly anyone else uses
raytracing because it is so completely wasteful. SGI uses a scanline based
rendering algorithm with Gouraud shading and Phong lighting. In this method,
each polygon in the scene is rasterized (conerted to a bitmap raster) in a
scanline fashion across its 2D view surface (no beams are traced). Scanline
algorithms are generally more complex than tracing and frequently use various
tricks to achieve an appealing result, but are usually magnitudes faster than
full raytracing. Reflections are accomplished through reflection mapping and
are not at all accurate (but visibly can look pretty darn good when done
right). Transparency is accomplished a number of ways including defining an
"alpha" opacity value for each polygon which gets taken into account when the
pixels of that polyon get blended with the background. Shadows can be
accomplished using z-buffer techniqes or may be faked with transparent
polygons. So to re-iterate what Jamie told you, SGI does NOT do raytracing.
%~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~%
% ` ' Mark Thompson CONCURRENT COMPUTER %
% --==* RADIANT *==-- ma...@westford.ccur.com Principal Graphics %
% ' Image ` ...!uunet!masscomp!mark Hardware Architect %
% Productions (508)392-2480 (603)424-1829 & General Nuisance %
% %
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Robert Gentner
if there are any 060 accelerators in the works. If anyone knows anything about
the company, specs, date of availability, or cost of such a product I would
appreciate an info you could provide.
Thanks in advance.
_/_/_/ ""\ ""\ ""\ ""\ """""""\ ""\ _/_/_/_/_/_/_/_/_/_/_/
_/_/_/ """"\ """\ """\ ""\ ""\ """"\ _/ gentner _/
_/_/_/ ""\ ""\ ""\"""\""\ ""\ ""\ """\ ""\ ""\ _/ @anasazi.com _/
_/_/_/ _/_/_/ """"""""\ ""\ "\ ""\ ""\ ""\ ""\ """"""""\ _/ Phoenix,AZ _/
_/_/_/_/_/ ""\ ""\ ""\ ""\ ""\ """""""\ ""\ ""\ _/_/_/_/_/_/_/
_/_/_/ Amiga is Important! Commodore is Irrelevant!
Stanley Anderson
>Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
>if there are any 060 accelerators in the works. If anyone knows anything about
I'm not aware of anything beyond the 50mhz 030's.
Stan Anderson
Joel Corn
: Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
: if there are any 060 accelerators in the works. If anyone knows anything about
: the company, specs, date of availability, or cost of such a product I would
: appreciate an info you could provide.
And who supposedly has an 040 accelerator for the 1200? This is news to
me.
Joel
--
*****************************************************************************
* Joel A. Corn - Owner/Head Programmer dark...@golden.ncw.net *
* DarkSoft Computers/DarkSoftWare 1-509-886-0581 *
*****************************************************************************
Roger Uzun
Robert Gentner <gen...@anasazi.com> wrote:
>Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
>if there are any 060 accelerators in the works.
I have never even heard ofplans for a 68040 accelerator for the A1200,
are you sure there are any available?
-Roger
--------------------------------------------------------------
bix: ruzun
NET: uz...@crash.cts.com
*WINDOWS from the folks who brought you EDLIN
Jason Weber
>Ours did. Also, take a look at the "Ideas in Motion" demo. You will see
>the SGI logo rendered, casting a shadow on the ground (with light
>emanating from a fixed source, which has the shape of a lamp). Not only
>does it produce shadows, even the surface of the logo is very plastic. In
>other demos the computer calculates transparency in real-time as well,
>with light sources of different color changing the color of the shadows,
>etc. If you don't call this ray tracing, I don't know what is. Of course
>it doesn't look anything like Lawnmower Man, but raytracing does not have
>to be complex to be raytracing.
Those shadows are just a unique application of texture mapping. They
give you the source of a shadow demo in 4Dgifts. They've had transparent
polygons for eons. You just have to be sure to manually depth sort
all your poylgons. In addition, the RE2 can anti-alias and use
12-bits-per-component color without loss in performance.
The point is that polygonal rendering can do just about everything
ray-tracing can do, just a whole lot faster. You can even use texture
mapping for limited reflections.
BTW, rumor has it that SGI's next engine should be about 10 times faster.
_ Jason Weber Phone:(703)704-1796
( \ _ \ /_ / _ _ Teletronics International 704-3901
\|(\/)())) \/\/(-/_)(-/( Army Research Lab Fax: 704-3196
// Modeling & Simulation Branch
(/ ja...@belvoir-arl-irisgt.army.mil OR @nvl.army.mil
Pete Rittwage
>Thanks in advance.
I didn't realize they made an 040 for the 1200 yet! I thought
it ran way too hot to do it safely.
Pete Rittwage
bush...@netcom.com
Chris J Schnurr
>Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
>if there are any 060 accelerators in the works. If anyone knows anything about
Please provide info on the 040 accelerator you're considering for the A1200.
I'm not aware of anything beyond the 50mhz 030's.
Stan Anderson
Please provide info on how you are going to stop your 040/060 from turning your
computer desk into a very large heat sink, melting any hardware that might be
attched in the process... ;)
Chris
cee...@uk.ac.hw.cee
Triumph through two stroke technology... Vor Sprung durk Trabbi
u898...@csdvax.csd.unsw.edu.au
> Robert Gentner (gen...@anasazi.com) wrote:
> : Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
> : if there are any 060 accelerators in the works. If anyone knows anything about
> : the company, specs, date of availability, or cost of such a product I would
> : appreciate an info you could provide.
>
> And who supposedly has an 040 accelerator for the 1200? This is news to
> me.
There exists a 68040 accelarator fro the A500 but whever would buy such
a product is beond me. Even the A530 from GVP would make your A500 cost more
than an equivalent A3000 !!
>
> Joel
George
Per Espen Hagen
> gen...@anasazi.com (Robert Gentner) writes:
>
> >Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
> >if there are any 060 accelerators in the works. If anyone knows anything about
> >the company, specs, date of availability, or cost of such a product I would
> >appreciate an info you could provide.
>
> it ran way too hot to do it safely.
I've also never heard of an 040 board for the 1200, and I can't imagine there
will ever be one -- even with a 50 MHz 030 it gets VERY hot in there.
With an '040, you'd have to install both a fan and a huge heat sink. And get
a much bigger power supply, possibly also replace the power wiring to the
expansion board... And the board alone would probably cost as much as an
A4000/030.
Maybe the original poster has read about the "A1240" board (DKB?). That's
*NOT* a 68040 board, it's just a piece of stinky marketing. The '40' stands
for 40 MHz; it's just another 68030 board...
A 68060 board, OTOH, now _that_ might be possible... The 060 is supposedly
a low-power unit. Relatively speaking, of course.
--
Per Espen Hagen per-esp...@ffi.no Tel: +47 63807653 //
Senior Scientist Image Processing Group NDRE, Norway \X/
Any resemblance between the above views and the views of my employer,
myself, or the view out of my window, is non-deterministic.
Alan L.M. Buxey
: > And who supposedly has an 040 accelerator for the 1200? This is news to
: > me.
: There exists a 68040 accelarator fro the A500 but whever would buy such
: a product is beond me. Even the A530 from GVP would make your A500 cost more
: than an equivalent A3000 !!
the A500 had that 040 accelerator from PP&S because it was large enough
to house it.
The trapdoor in the A1200 would be too small for the heatsink required
-> unless it had a very thin fan cooling system.
BTW, the A530 doesnt make the A500 more expansive than an A3000.
And, WHERE is this 040 A1200 card! :)
alan
From Alan, replies appreciated!___ __ _ __ ___ _
.----------------------. ///\\ |\\ /| || // ` /\\ __ __ 32
| Alan Buxey | __ /// \\ | \\ /|| ||(( __ / \\ // ||\\
|kc...@uk.ac.susx.solx1| \\\///--- \\l \/ ll ll \\_ll/--- \\ \\_ ||//
`----------------------' \XX/Amiga - The Reason That God Created Silica!
Aaron Smith
CS> Please provide info on how you are going to stop your 040/060 from turning
CS> your
CS> computer desk into a very large heat sink, melting any hardware that might
CS> be
CS> attched in the process... ;)
Wow that was funny.. anyway, the 040 is hot yes.. but the 060 isn't
that hot.... drop a msmall fan or heatsink on it and I don't think you have to
worry.
ahhh now imagine a small computer like the A1200 with 1.5 times the
speed of a pentium.. that would be interesting... Cirtainly worth a mention in
a few computer mags...
" Today Commodore UK/Samsung/phillips/whoever demo'ed a Amiga 1200
computer not a whole lot bigger than a Large IBM keyboard running the 66Mhz
68060 processor. This little computer has more overall power and Multitasking
abilities than a 66Mhz Pentium machine costing well over $2000. C= has
announced thier new modem will run a mere $950 street price"
ok a dream, but a fun one!
-Mustang Let's Rock.
life; like a story we wish was neverending,
always hoping there's one more page to go.
-Phil Collins
Eyal Teler
|> Please provide info on how you are going to stop your 040/060 from turning your
|> computer desk into a very large heat sink, melting any hardware that might be
|> attched in the process... ;)
Isn't the 060 a cool chip? (In more than one way :)
ET
Jeff Hanna
Chris J Schnurr <cee...@cee.hw.ac.uk> wrote:
>
>Please provide info on the 040 accelerator you're considering for the A1200.
>I'm not aware of anything beyond the 50mhz 030's.
There aren't any 040s for the 1200. At one point GVP was rumoured to be
working on one that would've used the low-power (.3 micron) version of the
chip. I dont' think this low-power 040 ever surfaced, though.
>Please provide info on how you are going to stop your 040/060 from turning your
>computer desk into a very large heat sink, melting any hardware that might be
>attched in the process... ;)
>
This would only be a problem with a true (or LC) version of the 040. The
060 will be a .3 micron chip by default with a much lower heat production
than the 040. Combine that with the fact that the 060 is actually closer to
the 030 address- and instruction-wise (although still higher performance than
the 040) and the 060 becomes a really good choice for an A1200 CPU
upgrade.
-------------------------------------------------------------------------------
Jeff Hanna | "Albert Einstein nailed space-time, but the
quar...@expert.cc.purdue.edu | Wild Thing had him stumped."
quar...@genie.geis.com | -Thomas Dolby
-------------------------------------------------------------------------------
Maxwell Daymon
: In article <CtK18...@cee.hw.ac.uk>,
: Chris J Schnurr <cee...@cee.hw.ac.uk> wrote:
: >Please provide info on the 040 accelerator you're considering for the A1200.
: >I'm not aware of anything beyond the 50mhz 030's.
: There aren't any 040s for the 1200. At one point GVP was rumoured to be
: working on one that would've used the low-power (.3 micron) version of the
: chip. I dont' think this low-power 040 ever surfaced, though.
Yep. It did. GVP isn't very interested in the Amiga anymore. (AFAIK)
It's called the MC68040V
--
// | "NEVER EVER mess with a PCB jumper you don't
// Maxwell Daymon | understand, even if it's labelled 'SEX AND
\\ // mda...@rmii.com | FREE BEER'."
\X/ | -- Dave Haynie
Jason Compton
>On Tue 26-Jul-1994 3:48p, Chris J Schnurr wrote:
>
>CS> Please provide info on how you are going to stop your 040/060 from turning
>CS> your
>CS> computer desk into a very large heat sink, melting any hardware that might
>CS> be
>CS> attched in the process... ;)
>
> Wow that was funny.. anyway, the 040 is hot yes.. but the 060 isn't
>that hot.... drop a msmall fan or heatsink on it and I don't think you have to
>worry.
A fan? Where's it going to blow the air to? There's really not enough room to
do any good.
Jason Compton jcom...@bbs.xnet.com
Emulation Editor Amiga Report Magazine
Editor-In-Chief Amiga Report Coverdisk
Chief Executive Officer DigiForte Designs, Ltd.
FAX: (708) 741-0689
The time to rise has been engaged. -REM, Finest Worksong
Robert Gentner
>: Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
>: if there are any 060 accelerators in the works.
>And who supposedly has an 040 accelerator for the 1200? This is news to me.
>
>Joel
Of course there isn't an 040 accelerator for the 1200! As we all know, it would
melt the A1200 around it. At the time I posted this, I was dreaming of a 100
mips 060 A1200 and lost touch with reality (That and the fact that I had just
finished a 12 hour shift). Sorry for the error it was supposed to be '030'.
--
Marco Nelissen
: In article <30vdqi$c...@golden.ncw.net> Joel Corn writes:
: >: Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
: >: if there are any 060 accelerators in the works.
: >:
: >And who supposedly has an 040 accelerator for the 1200? This is news to me.
: >
: >Joel
: Of course there isn't an 040 accelerator for the 1200! As we all know, it would
: melt the A1200 around it. At the time I posted this, I was dreaming of a 100
: mips 060 A1200 and lost touch with reality (That and the fact that I had just
: finished a 12 hour shift). Sorry for the error it was supposed to be '030'.
There IS a 68040 card for the 1200. The kit comes with some things to lift
your 1200 about 5 centimeters (that's two inches, guys ;), so that the fan
that is placed underneath the unit can cool the CPU. It's available here
in holland (forgot the company's name, will look it up next weekend), and
I read a German company is going to release an '040 card as well (forgot
their name too, will look that one up tomorrow).
.
I just read a (p)review of a 68060 card for the 3000/4000 BTW. The arcticle
showed a screenshot of SysInfo, which indicated a speed of 2.04 times that
of a A4000/40, and the comment was 'Phone me NOW!!!'.
I must say I am a bit disappointed in the speed. OK, it's twice as fast, but
it gets twice as much MHz, so that's logical. How about al those
architectural improvements that are supposed to be stuffed into the 68060?
How come it isn't at least 3 times as fast?
.
Marco
.
(no fancy signature yet)
Ralph Schmidt
>I just read a (p)review of a 68060 card for the 3000/4000 BTW. The arcticle
>showed a screenshot of SysInfo, which indicated a speed of 2.04 times that
>of a A4000/40, and the comment was 'Phone me NOW!!!'.
>I must say I am a bit disappointed in the speed. OK, it's twice as fast, but
>it gets twice as much MHz, so that's logical. How about al those
>architectural improvements that are supposed to be stuffed into the 68060?
>How come it isn't at least 3 times as fast?
They haven't activated everything of the 060 yet...with all on you can 2x
the results. With the little Mips program available and all activated you
get about 83Mips and the A4000/40 gets 17.
Please no discussion about Mips is worthless or is the dead end of speedtests.
The reason i can only tell the Mips value that indicates the speed difference
because AIBB 6.1 doesn't run. It's using supervisor commands the 060 doesn't
support->Nirvana.
--
Ralph Schmidt la...@uni-paderborn.de
University of Paderborn (Germany)
Joel Corn
In article <1994Jul31.0...@anasazi.com> gen...@anasazi.com (Robert Gentner) writes:
> In article <30vdqi$c...@golden.ncw.net>dark...@golden.ncw.net(Joel Corn) writes:
> >: Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
> >: if there are any 060 accelerators in the works.
> >:
> >And who supposedly has an 040 accelerator for the 1200? This is news to me.
> >
> >Joel
>
> Of course there isn't an 040 accelerator for the 1200! As we all know, it would
> melt the A1200 around it. At the time I posted this, I was dreaming of a 100
> mips 060 A1200 and lost touch with reality (That and the fact that I had just
> finished a 12 hour shift). Sorry for the error it was supposed to be '030'.
Well, if they used the low voltage 040 it wouldn't be too bad.
However, an 060 would be the dream machine! I currently have the 50
MHz CSA 12 Gauge and heartily recommend it over anything else. I
previously had a 33 MHz Microbotics board, but it didnt't autoconfig
the memory like the 12 Gauge does. The M'botics board DID have a clock
tho.
And stay away from the GVP garbage and their proprietary SIMM's.
You'll save MUCH money getting either the M'botics or CSA board and
using the industry standard stuff than GVP's $1200 16 meg SIMM's. My
16 meg SIMM cost me under $600. Quite a diff, huh?
Joel
--
**********************************************************************
* Joel A. Corn - Owner/Head Programmer - DarkSoft Computers *
* email:dark...@golden.ncw.net - phone/fax:509-886-0581 *
**********************************************************************
Lasse Olsen
: Robert Gentner (gen...@anasazi.com) wrote:
: : In article <30vdqi$c...@golden.ncw.net> Joel Corn writes:
: : >: Before I buy an 040 accelerator for my A1200, I was wondering if anyone knows
: I just read a (p)review of a 68060 card for the 3000/4000 BTW. The arcticle
: showed a screenshot of SysInfo, which indicated a speed of 2.04 times that
: of a A4000/40, and the comment was 'Phone me NOW!!!'.
: I must say I am a bit disappointed in the speed. OK, it's twice as fast, but
: it gets twice as much MHz, so that's logical. How about al those
: architectural improvements that are supposed to be stuffed into the 68060?
: How come it isn't at least 3 times as fast?
The AIBB and SysInfo tests I've got from Phase Five showes the CyberStorm
060/50 beeing 3-4 times as fast in integer and 5-6 times as fast in FPU
as the A4000/040. The memory handling is over 10 times as fast due to the
crippeled A4000 cpu-board design.
Christian Bauer
Pete Rittwage wrote in comp.sys.amiga.hardware about "Re: Any news on an
060 for A1200?" this:
PR> I didn't realize they made an 040 for the 1200 yet! I thought
PR> it ran way too hot to do it safely.
Yes, that's right. Excuse me, but I can't remember, that I've read
anything about 68060. Could someone send me Information via PM ?
But please note my new adress : C.B...@fsw.hn.bawue.de !
Thank you !
--
Christian Bauer