7 Years of 3D Graphics

xtra writes "At Accelenation they are running a nice timeline about 7 years of pc 3d graphics contains much info and even talks about some of the not so well known players anyone still remember rendition? or BitBoys?" How many cards on their timeline chart have you used?

Comments

Only 7 years?

[Thnurg]

I was playing Elite in 1984. Damn, that was a fine game.

Re:Only 7 years?

[Bilestoad]

Elite lives on - if you have a PDA there are three or four clones, a couple of them GREAT, right down to the 3D combat. When I did my knee on a snowboard trip a year ago, the Visor Prism with an elite-clone helped me forget about the snow.

Re:Only 7 years?

[tjwhaynes]

You mean Void? Or did you have another one in mind?

Cheers,

Toby Haynes

Infinity and beyond...

[ackthpt]

At the going rate, the board with CPU and chipset will be a daughterboard of the graphics motherboard. :]

Re:Infinity and beyond...

[Telastyn]

Or perhaps better yet the motherboard will just be a backplane for Graphics, Sound, Physics, AI, Network modules and the such.

Some machines might just need Math modules (all you SETI junkies.)

Though unfortunately computing for the home seems to be moving towards "all in one" motherboards and the such. :/

Re:Infinity and beyond...

[Boone^]

the GF3 uses vectors for calculations. An Athlon using the x86 CISC requires more overhead per instruction.

Athlon:
add ax,bx; % where ax and bx are 32-bit scalars

GF3:
add reg0, reg1; % where reg0 and reg1 are vectors of scalars

If you wanted to do the same work of the GF3 on an Athlon, you'd need 32 (or however deep the vector registers are) successive instructions.

Once you ditch the instruction overhead for doing an operation on X number of successive scalars, the processor spends more time doing the math and the FLOPS goes up. Take a look at the AltiVec unit in the G4, or CRAY vector supercomputers.

Re:Infinity and beyond...

[PurpleFloyd]

Take a look at the nVidia nForce (linked to a Tom's Hardware review) boards. They have an integrated Geforce 2 GPU, DDR support, 5.1 Dolby Digital sound, and seem to be priced fairly reasonably. They don't have the performance of a real high-end system, but they do have decent upgradablity and fairly good 2D and 3D performance (at least compared to the old ATI chips that seem to be common on all-in-one motherboards).

Why 7 years?

[Jacek Poplawski]

Why only 7 years of 3D graphics on PC?
What about Stunts, Elite, and other 3D games?

Re:Why 7 years?

[Strog]

From the article:This article should really be called "7 Years of 3D Graphics" because 3D acceleration seemed a suitable starting point..

The point was accelerated 3D not "software" 3D. I still remember bombing runs from Ace of Aces on a Commodore 128 though it definitely wasn't a hardware accelerated game.

Re:Why 7 years?

[Rogerborg]

• What about Stunts, Elite, and other 3D games?

Given that ZX81 (aka TS1000) emulators for the PC were around from day dot (or a ZX81 emulator running on an Atari ST emulator running on a PC...) how about 3D Monster Maze? It's not that much more primitive than Doom. Run! Run from the scary Tyrannosaur!

(Yes, I know, the article is probably about hardware that draws triangles real fast, but it's Slashdotted hard, so we may as well have some fun reminiscing. If nothing else, it'll confuse the young 'uns ;-) )

Re:Why 7 years?

[CaseyB]

I still remember bombing runs from Ace of Aces on a Commodore 128 though it definitely wasn't a hardware accelerated game.

It wasn't even 3D. Sublogic's Flight Simulator and "Jet" were 3D, which is why they ran at about 1fps on a C64!

Re:Why 7 years?

[Lussarn]

The point was accelerated 3D not "software" 3D.


The amiga had filling of polygons in 1985 using the blitter. $DFF058 was the register use to fire up that beast if I remember correclty.

Re:Why 7 years?

[printman]

But there have been accelerated 3D boards for a lot longer - I was programming some boards from SubLogic (remember them?) and then from SGI (IRISvision - Personal IRIS graphics on a PC with a *blazing* 14k polygons/second with hardware Z-buffer! :)

The article might better be titled "7 Years of *Consumer* 3D Graphics Cards"...

Re:Why 7 years?

[CaseyB]

do you not recall STUNT CAR RACER?

Hmm. Release year 1989. Nope, after my time!

or how about the mystical SENTINEL?

I had a copy of that one. I recall it was one of very few games that I couldn't figure out how to play by just playing around pressing all the keys. So no *good* memories, anyway.

How fast do we really need to go?

[TrollMan 5000]

Can graphics technology possibly get any faster? Well the GeForce2 GTS chip ran Quake3 at 80fps in May of 2000. Just twenty-two months later a GeForce4 Ti4600 can run Quake3 over three times faster. On that reckoning the GeForce6 in two years time should be running Quake3 at over 700fps. Is that fast enough for you!

Is there really much visual difference between 700 fps and 135 fps? I'm not really sure if the human eye can make the distinction. They're sure pretty-looking numbers, but do the results show for it?

And how long before video cards can render essentially photo-realistic graphics? Soon games will be more like interactive movies.

Re:How fast do we really need to go?

[alen]

It's not the speed anymore, but the visual quality that matters. The Geforce4 can play games in 1600 by 1200 resolution with all effects maxed out at decent frame rates. Anything past 60 FPS doesn't really matter to a lot of real game players anymore. What matters is the visual quality of the game. How real does the smoke look? How real do the characters move? etc.

Re:How fast do we really need to go?

[Fweeky]

> Is there really much visual difference between
> 700 fps and 135 fps?

1. 700FPS means when things are really hectic you drop to 300FPS instead of 30FPS (this is basically the argument for 150FPS over 70FPS; the average is fine, but the most detailed parts will start pushing it into being jerky)
2. 700FPS means you can up the detail ~7 times and get ~100FPS. Unreal 2 is already pushing about 4 times as many polygons as the orignal engine; you may well get 700FPS now, but you certainly won't in the top end games in 2 years time.

Remember, Quake 3 is fairly old now; already games like MoH have parts that will make most above-average machines struggle (like that mission with all the trees); newer engines, larger cheaper monitors etc are only going to push that further.

Re:How fast do we really need to go?

[SirSlud]

For whatever reason, in Quake3, having > 200 fps allows you to jump to a height a tiny bit higher than having 200fps (probably because you actually hit the 'peak' of the jump in a frame instead of 'between frames' (ie, the peak of your possible jumpable height happens on a frame instead of a point interpolated between frame, so the physics engine picks up on this and lets you up there.))

So it may not help the visuals, and aside from using the extra frames for motion blur, etc, it also provides you physics engine with a more 'correct' version of your path .. although this is probably dependant on how one implements their physics engine.

Re:How fast do we really need to go?

[_UnderTow_]

Sound like flamebait but I'll reply anyway.

Granted, you won't notice any difference past your monitors refresh rate. But there are a lot of reasons to have a much higher frame rate. When you do a benchmark you are getting the AVERAGE framerate, there are times when the actual framerate is much lower than this. I figure if I can keep my average framerate above 100fps then it shouldn't dip below my refresh rate.

One other things is that you might be able to get 200-300fps in quake3 right now, but some benchmarks of the new Unreal engine on anandtech show that those same cards that are spanking the q3 engine will be spanked by the new unreal engine. And I don't imagine the new Doom engine will be any easier. Having extra horsepower now helps when the newer games come out.

Re:How fast do we really need to go?

[ergo98]

Is there really much visual difference between 700 fps and 135 fps? I'm not really sure if the human eye can make the distinction. They're sure pretty- looking numbers, but do the results show for it?

Quake 3 just happens to be a benchmark (and an old one at that) whose numbers are relative, but not necessarily realistic : Imagine if they benchmarked harddrives by always saying "The IBM 75GXP can load 40,000 10KB hello.c files / second", to which everyone follows that up by commenting that they only need to load 1 hello.c file, etc. In other words, for a more demanding task like the new Doom, Quake 3 with complex mods like Urban Terror, or much more complex games like Operation Flashpoint, 135fps in the stock Q3 equals ~15 fps in a complex outdoor scene in OpFlash. And as has been recapped many times in the past: We are just touching the surface of realistic environments (i.e. try to model nature in a dynamic fashion and the best boards put out single digits FPS, if that).

Re:How fast do we really need to go?

[Hieronymus Howard]

Ray tracing isn't the only way to achieve high-quality graphics. Renderman (used for Toy Story, Monsters Inc. and plenty others) isn't a ray tracer, but a sub-pixel renderer (if I remember my computer graphics M.Sc. correctly).

We've still got a very long way to go until we get Monsters Inc. quality real-time games. As you say, current cards render triangles. Curved surface rendering (e.g. NURBS) may come next. Anti-aliasing takes a lot of power. I think that current cards are still using Gourard shading, which is the most primitive shading model there is (correct me if I'm wrong here). The next step is Phong shading for highlight effects (there are hardware-optimised Phong shading algorithms, but they're still slower than Gourard). Then there's deformation mapping (Renderman again), etc. etc.

I believe that Quake 3 etc. does use radiosity algorithms, but that doesn't need to be done in real time, just when the level is compiled.

HH

Re:How fast do we really need to go?

[EllF]

The "SOFTWARE realtime raytracing" you're referring to is rendering extremely low-detail primitives (spheres, cones, etc.) in one-pass tracing at a tiny resolution, then bilinear filtering the image and scaling it. This is neither photorealistic (as the original poster was inquiring after) nor impressive.

What was being discussed was not 1x1 traced, static geometry localisation demos with environment mapping and scanline rendering. What was being discussed was full-fledged, true real-time raytracing of complex scenes with lighting and reflection taken into account in the trace passes. Show me something with even the complexity of Doom in a real-time raytraced environment, I'll be impressed. Half-way demos like this (not that they aren't pretty) just don't cut it.

Re:How fast do we really need to go?

[bribecka]

The problem is that right now, we're rendering triangles. We can do an incredible number of them - most cards list their triangle output rate - but it just isn't going to be good enough. The holy grail of 3D graphics, in my opinion, is real-time ray tracing.

But even in raytracing, the overwhelming majority of objects are made up of triangles. Besides quadrics, most of the things are just converted to triangles anyway.

With a ray-traced image, shadows, reflection, and refraction are accurate and free (no extra CPU time needed), but the rendering itself takes a while.

While these effects do somewhat come "for free", reflection and refraction do take extra CPU time. For each reflection/refraction point, you have to shoot another ray. It's free in the respect that it's the same algorithm, there isn't any fancy special-case nonsense to get these effects. But, for example, if the original ray hits an object that is both reflective and translucent, two more rays are shot, and each of those rays need to interact with other objects, possibly shooting more rays. Try rendering a scene with a tracing depth of 1, then try it with a depth of 10. Big difference.

It's a ways off, but I'd guess that twenty years might see such technology within the reach of wealthier consumers.

Man, I hope it's not that far off...really, if you think about how far things have progressed even in the past 3 years, it's not hard to imagine Toy Story-like graphics in games in just a few years.

Re:How fast do we really need to go?

[GigsVT]

And by the time the newer games come out, you could have spent 50% less for the same hardware.

Re:How fast do we really need to go?

[bribecka]

Renderman (used for Toy Story, Monsters Inc. and plenty others) isn't a ray tracer, but a sub-pixel renderer (if I remember my computer graphics M.Sc. correctly).

You're referring to PhotoRealistic Renderman (PRMan), the actual product developed by Pixar. It uses the REYES algorithm.

RenderMan is a specification for defining 3D scenes, much as PostScript is a specification for defining 2D documents/images. There are many renderers that are RenderMan compatible, including raytracers such as BMRT.

Re:How fast do we really need to go?

[John_Booty]

Try rendering a scene with a tracing depth of 1, then try it with a depth of 10. Big difference.

I suppose letting the user adjust the raytracing depth would provide an easy way to let the user adjust the balance between frame rate and graphical detail.

Current games usually have about 20 different variables you can tweak to achieve the same result, but it's often hard to tell which variables are actually having an effect, let alone what some of them do!

Re:How fast do we really need to go?

[SuiteSisterMary]

Here's the thing. Film runs at 24 FPS, but film also captures motion over time. Lets say you have film running at 2 FPS. You record a hand waving across a table for one second. Your two frames will show a blur of a hand moving across the table. Now, make a 3d engine that renders that hand moving across that table for 2 FPS. You'll see a static picture of a hand at one side of the table, then a static picture of the hand at the other side of the table. In other words, film runs at 24 fps, but it captures more information over time in a single frame. Computers, however, render static images. And therefore you need more static images per unit of time to simulate the 'exposure' quality of film. This is why 3dfx went for the T-buffer; this would allow for high FPS style imagery with fewer actual FPS.

Re:How fast do we really need to go?

[CaseyB]

But there are a lot of reasons to have a much higher frame rate.

Wrong. Excessive framerate is pointless.

What you mean is "It's good to have the power capable of rendering 700 fps in benchmark X, because that much power will allow you to maintain the optimum framerate in all circumstances.".

Re:How fast do we really need to go?

[Fweeky]

> Framerates above the refresh rate of the monitor are pointless.

While you're right that there's little point in rendering faster than the monitor refresh beyond benchmarks (which is really what we're talking about), being able to go above the refresh rate on average scenes is important because complex scenes will be, um, more complex and drop the framerate.

You may be able to push an average of 90FPS in Quake 3, but that's not so wonderful if you drop down to 20FPS every time you leave a corridor and drop to 10FPS every time someone throws a smoke grenade or fires a rocket.

Re:How fast do we really need to go?

[Wolfier]

Can't agree more.

Although, from a bandwidth perspective hardware tesselation will be a huge step.

If we could describe a sphere as a sphere instead of 300 triangles, an arc as an arc instead of 200 triangles, then who needs AGP graphic cards, or even PCI for that matter?

Re:How fast do we really need to go?

[Wolfier]

Theoretically, yes. But how do you define "dense enough"?

Experiments and researches have shown that (take a look at some SIGGRAPH 99-01 papers) even in high triangle count, Gouraud shaded objects still look inferior to the same Phong shaded object using fewer triangles.

And the bomb is, texture-based Phong shading is a relatively simple technique to implement, gives good results if you have good texture hardware, and fast.

Even with T&L I highly doubt a Gouraud shaded object with "dense enough" triangles to look on par with the same Phong shaded object with fewer triangles would render faster. (in fact I'll bet you it'll render slower)

Re:Number Nine

[ackthpt]

No mention of Tseng Labs. Maybe the 4000 chipset wasn't considered a Graphics Accelerator, but having one of those in a new 386DX25 was pretty cool and what I first played Return to Castle Wolfenstein and a few ID games on.

Only 7 Years?

[green pizza]

Heck, I still remember the "which is better, Silicon Graphics Reality Engine or Ferrari Testarossa?" threads in the USENET from the summer of 1992. Even the dual pipe / dual head SGI VGXT "Skywriter" from 1989 was pretty damned impressive. Even many, many years later.

Re:Number Nine

[Jeffrey Baker]

Uh? I don't remember any #9 cards with 128MB of memory. In fact, #9 was out of business two years ago. You may remember the #9 i128 series of cards, but those are very old and do not have 128MB memory.

You only need 16MB to handle th highest resolution computer graphics displays ever made.

Rendition? Sure I remember...

[dreadpiratemark]

The Rendition cards, like most cards, were quite good presuming that the folks coding the games knew how to utilize the power of the card. My personal favorite game (Grand Prix Legends) is a little over 3 years old but was written specifically for Rendition cards and Voodoo2 cards - OpenGL support was tacked-on later. And of the three, there was no doubt that Rendition was the best for that game. Nothing inherantly wrong with the cards, imo, but they just got done in by the Voodoo cards of the time (which, of course, got done in themselves not much later).

Anyway, thanks for asking if I remembered Rendition!

Cheers.

Demo scene.

[saintlupus]

In a related note, what the hell ever happened to the Future Crew? Man, I remember waiting with bated breath for Second Reality to download over my 2400 baud modem.

--saint

Re:Demo scene.

[2Flower]

Future Crew reassembled, broke up, faded out, tried to get into the games biz, etc. Their site has been offline for awhile.

Fortunately the demo scene lives on; pouet hosts links to nearly every demo in existence across multiple platforms. And to keep us on topic, most demos nowadays are 3-D accelerated. It's become less a game of "What techie tricks can you do?" and more a game of "How artistically can you use the technology?". There's some visually striking demos being made nowadays, and not just because they have shadebobs or glenz cubes.

Re:Demo scene.

[don_carnage]

Wow...haven't thought about Second Reality in forever. I remember getting a new Diamond video card and immediately playing Second Reality to see how much faster it ran -- it was my benchmark.

Ok...anyone remember Kosmic Free Music Foundation's "Little Green Men"? What about good old Cubic Player and .mod's?

Re:Demo scene.

[don_carnage]

How artistically can you use the technology?"

For me it was the music in sync with the visuals. A demo can have the coolest visuals ever, but if the tunes suck...

Re:Demo scene.

[sph]

In a related note, what the hell ever happened to the Future Crew?

The actual Future Crew is no more, but many of the members have been active in various projects in for example gaming industry. You may have heard of Max Payne, made by Remedy Entertainment, or 3DMark, made by MadOnion. Though not really related to FC, they both employ former FC members, and may be the best known examples. As for other demosceners, some Byterapers members were involved in Rally Trophy, made by Bugbear. It also features some music by Purple Motion/FC.

Any other examples of demosceners, perhaps from outside of Finland?-)

Re:Demo scene.

[Cruciform]

Skaven did the music for Bejewled (Windows Version).

Palm version here, but I'm guessing it lacks the tunes :)

Slashdot new Slogan

[Zapdos]

History for Nerds.

Just seems to be allot of history stuff lately.

Two words :

[Viking Coder]

"Motion blur".

Or how about "rendering passes"?

Or how about "anti-aliasing"? (Kind of cheating on that one.)

Or how about "soft shadows"?

In short, more is better. If you give me higher framerate, I'll figure out what to do with those extra cycles.

Re:Two words :

[Viking Coder]

I forgot to mention "reduced latency."

If you can render the entire image in the 1/700th of a second before the screen refresh, you can much more accurately track events than if it took you 1/60th of a second to render the image. USB mice allow a much higher sampling rate that you could take advantage of, for instance. When you incorporate things like motion-tracking polhemus devices, and a head-mounted display, things get really interesting.

Re:Two words :

[CaseyB]

There's no point in having a framerate any higher than the refresh rate on your display

Your points have nothing to do with framerate. Yes, more power for more visual quality is better, but a higher number of frames per second makes ZERO difference, once you surpass the refresh rate of the monitor. In fact, it's worse, since you're wasting resources that could be allocated to other parts of the system.

Re:Two words :

[Viking Coder]

Motion blur is accomplished by rendering multiple frames, and then combining them. One method of anti-aliasing is to render multiple jittered frames, and then to combine them. Soft shadows can be done the same - multiple renderings with jittered light sources, and then combining them.

The rate of preparing frames and the rate of displaying frames are related, but saying that additional framerate beyond the refresh rate "makes ZERO difference" is incorrect.

Also, my latency point holds quite well. Framerate is the rate at which you may draw frames, wouldn't you agree? If you have an increased framerate, you may either draw more frames than you render (which is foolish, unless you combine them, somehow - as in motion blur, anti-aliasing, soft shadows), or you can simply delay drawing the frame until the last possible instant before you need it. Doing so will decrease the perceived latency, especially if you have a higher sampling rate for your input devices than your refresh rate on your display - which I believe is quite possible with USB mice, and I'm certain it's possible with firewire devices.

To borrow an analogy, framerate is the speed of your car. If you only intend on driving a certain overall speed (a given refresh rate), there are other ways you can use that extra speed. You can travel a greater overall distance (motion blur, multiple rendering passes, anti-aliasing, soft shadows), or you can simply leave later and still arrive on time (for decreased latency.)

By the way, I'd like to analyze your term, "power for more visual quality"; "Power" is the rate of doing work. Therefore it sounds to me like what you're describing is "framerate," the rate of doing the work of preparing images. My analysis is that you're saying that "higher framerate is better," which was my original point.

Re:Two words :

[CaseyB]

Framerate is the rate at which you may draw frames, wouldn't you agree? ... Therefore it sounds to me like what you're describing is "framerate," the rate of doing the work of preparing images.

Framerate is the number of completed frames per second that are written to the framebuffer. Hence, any excess beyond the refresh rate of the monitor make no difference whatsoever to the viewer. The intricacies of the internal rendering procedures are irrelevant to the measure of framerate.

you may either draw more frames than you render (which is foolish, unless you combine them, somehow - as in motion blur, anti-aliasing, soft shadows), or you can simply delay drawing the frame until the last possible instant before you need it.

You're mixing your terminology. If you "draw more frames than you render", framerate is a function of how many times per second you "render". If you "delay drawing the frame until the last possible instant", the framerate is a function of how many times per second you "draw the frame".

My analysis is that you're saying that "higher framerate is better,"

Sorry, no.

Re:Two words :

[Viking Coder]

Why must a framebuffer only contain "completed frames"?

The proposed OpenGL 2.0 specification supports more arbitrary blending modes under the title "Frame Buffer Operations". By their definition, the framebuffer does not only store "completed frames," so your definition of framerate is incorrect.

Why are the intricacies of the internal rendering procedure irrelevant? I thought that's what we were talking about. If you want to pretend that rendering is a black box, and then magically an image is painted on the screen, feel free - but don't pretend you understand "framerate."

Maybe you're correct - maybe I should use the term "render rate" to refer to partially-completed frames per second. But I can't understand why you'd even _hint_ that more "render rate" isn't better.

Re:Just incremental improvements

[green pizza]

My voodoo2 SLI still kicks all kinds of ass.

Amen, brotha... my only real PC at home is a PII/400 with a pair of Creative Labs Voodoo2 cards running SLI. Win98SE is stable (enough) for the few games and utilities I run on it. And 56 FPS in Quake2 and 41 FPS in Unreal is good enough for me.

Re:Number Nine

[Zurk]

Actually i think number 9 had the first 128bit graphics chip on board not 128 megs of memory.
again i could be wrong.

scene.org!

[antdude]

Don't forget http://www.scene.org as well :).

calling Dr. Jim Clark...

[green pizza]

... it's time for the next wave of 3D.

I love playing with the SGIs at work and I enjoy playing with the wizbang PCs that my roommates and I have, but to be honest, I'm really not that impressed with modern gfx accelerators. The original geForce was pretty neat, and SGI's last big leap (InfiniteReality in '95) was cool... but golly, things really haven't changed much since Clark and his gang from Stanford opened our eyes to 3D in '82.

We've gone from cabinets to cards to chips to a single chip. We've added some gfx extensions and now do multiple rendering passes to make things look prettier... but really, nothing has changed much in the recent years. It's smaller, faster, cheaper. Steady evolution... but so is the scum growing in my bathroom sink.
Please excuse me while I yawn.

Re:calling Dr. Jim Clark...

[PotatoHead]

I totally agree. We need another innovation right now.

3D could use some work. Very few applications and displays make good use of it.

Re:Number Nine

[Milalwi]

You only need 16MB to handle the highest resolution computer graphics displays ever made.

This is true for 2D displays, but when you start having double and triple buffering plus z-buffers it starts to add up. Then add the texture requirement and you can see why most newer cards have 64-128MB of memory on the cards.

Milalwi

Re:Just incremental improvements

[BrookHarty]

I was playing tribes for the longest time, and voodoo2 sli got about 60-70fps in tribes. My diamond 8 meg tnt card couldnt beat it. Then my GF2 could, my GF3 ti500 not only beats it, I can turn on 4x anti-aliasing, and it looks smooth. Thou the game is better on 3dfx, the engine only renders it perfectly it under glide.

I bought shogo in the old games area at the store, tried to play it, had to turn off zbuffer/wbuffer to play.

Kinda a shame, when your old games wont play on new hardware. Lucky I have a p233, voodoo2sli/tnt2/sb32awe/128 megs with win98 :) With real 3dfx voodoo drivers just for that "Wow" old feeling. Of course if I want really old, i go dig out my c64 or Amiga500. Currently, I just run those under emulation.

Rendition? Did you say Rendition?

[Rogerborg]

I still have nightmares about developing for the Rendition Verite 1000, which was a lovely graphics decelerator on anything faster than a P100. When we got our first batch of Voodoo 1's delivered, there was a brief but very ugly struggle to get our clammy hands on them. You ain't seen pathetic until you've seen geeks wrestling and squealing like stuck pigs over 4Mb graphics cards, let me tell you.

Question to anyone else who has developed 3D graphics: who did you find driving the demand? In our games house, there was a running battle between the programmers and the artists. Us code monkeys were forever on at the artists to cut down the polygon counts, but they kept trying to slip in models that were barely stripped down from the FMV sequences. In the end, we came to an equitable solution: they won, the game ran at 10fps, and all the programmers left.

I wonder how many other games were ahead of their time in that regard, and how many of them would be rescuable given cards that scoff at polygons and eat dozens of 256x256 textures before breakfast?

Re:Number Nine

[ram.loss]

I think you mean the first consumer level graphics card to use a 128 *bit* data path. I remember seeing them bundled with Dells in the back cover of Byte a few years ago.

Moore's law applied to 3D graphics

[brer_rabbit]

I was wondering if anyone had applied Moore's law to 3D graphics. A quick google search and...

http://www.3dlabs.com/product/technology/moo resla.htm

Unfortunately it's a company paper and very biased towards the 3Dlabs Wildcat. That, and it's a bit dated. Then I found a Microsoft Research pdf:

http://amp.ece.cmu.edu/ECESeminar/slides/Whitted /F 00_Whi tted_slides.pdf

it's an interesting read, but not 100% relevant. Anyone else have relevant info?

Re:Moore's law applied to 3D graphics

[GoRK]

I don't remember where exactly I read it, but basically, there was a statement that 3D Acceleration was outpacing Moore's law by a monumental amount such that Moore's law was totally inapplicable. Maybe I can dig something up...

~GoRK

Ah, 3d hardware. The beginning of ugly.

[ethereal]

Yep, after all the fanboys started demanding the games in 3d, and then the game companies turned to supplying them, the effective graphic quality of computer games plummeted, and has only now maybe reached the beauty that we had at the pinnacle of sprite-based games. Sure, you could only see one side of the monsters, etc., but they were good-looking monsters - none of these chunky triangular-looking things that didn't even have fingers, toes, etc. and were plastered with dim-looking repetitive textures.

3d is almost getting good enough that I can stand to look at it. But for a while there, it really made games look a lot worse, just for some undefined promise of realism that was never really satisfied until maybe recently - those early 3d games just looked unrealistic in different ways than the 2d ones had. It's like the gaming industry fired anyone with taste and just kept all the techs.

OK, I think I'm done ranting now.

primitive shading model there is (correct me if I'

[dpilot]

Lambert shading is more primitive, essentially flat.

More memory

[Traa]

> You only need 16MB to handle the highest
> resolution computer graphics displays ever made

you will allways need more memory (in 3D graphics accelerators), even if the display resolutions don't increase. Lets say we settle for a nice 2000x2000 ish display. Thats 4M pixels, at 32-bit is 16MB for the display.
At least double (32MB) but preferably triple (48MB) buffer this so you can create a new frame while the old one is being displayed. Then we need a Z-buffer (or W-buffer) to hold the depth values (24bit values) for each pixel, so we know what is in front of what, typically you might want to do some stencil effects to (8-bits, can be packed with the Z-buffer) that would be another 16MB. Now we have the basics for a 3D graphics display and are at 48-64MB.

But we are not done yet, now for some more interesting effects:
- Texture memory. Typically use the leftover graphics memory and swap the rest from host memory (but we don't like swapping, so preferably all textures should be in onboard mem) 2-64MB
- 2x Antialiasing (1 Backbuffer + 1 Z-buffer 2*2*size of display buffer) = 64MB (4x antialiasing = 256MB)
- Shadowbuffer (rendering into a kind of Z-buffer from the lightsource to create realistic shadows) 16MB
- Accumulation buffer effects like motion blur (very expensive, a good blur could take 4 to 32 frames) or depth of view could make us want another 4-32*16=64-512MB

I for one could easily use more then 1GB of onboard graphics memory.

Interesting trend.

[Frag-A-Muffin]

The first thing that jumped out at me was this interesting trend I see on the chart.

The companies that have long red lines (meaning the time it took for them to ship since their announcements, ie HYPE) are all gone! :)

The ones that kept a relatively consistent schedule are still around. Once again, a smart business plan wins, not super-hyped, non proven stuff. ;) (Yeah .. I know. I'm biased. Nintendo Forever! :)

(On a side note, I wonder how long the line would've been for the xbox! :)

Re:Interesting trend.

[athakur999]

The XBox was announced in March, 2000. The Gamecube was announced in August of the same year. Both were launched in November last year. So the XBox line would have been a whopping 5 months longer :)

Nvidea and ATI

[AndyChrist]

Note the downward trend in those bars for product life there. They may not be entirely accurate, but note how the ATI and especially Nvidea bars generally get shorter and shorter....

Heh. Could this have any factor in their success?

Where are the real physics engines?

[linzeal]

Physics specalized processors? Can anyone show some nice linkage for them? That sounds like the next step for games today complete and utterly lifelike physics engines instead of scripted crap. Would make mapping much easier as well, imho. I know of geomod tech from the people that did red faction and freespace but what else is out there up to and including programs or languages for astrophysics and geomorphology simulations?

Re:Where are the real physics engines?

[Telastyn]

The only reason I mentioned the engine is because I'd love to design/write (I'd rather design) games one day, and that seems like something that would be a great thing for games, rather than re-implimenting Newtonian physics over and over again, there would be something akin to openGL that would have preset functions for things.

The processor itself would likely be a little specialized to handle (x,y,z,vx,vy,vz) style location/speed vectors and the such.

The closest thing I've seen to it was something one of my Materials Science TA's wrote to show and simulate forces on beams.

Re:Where are the real physics engines?

[Shiny Metal S.]

The processor itself would likely be a little specialized to handle (x,y,z,vx,vy,vz) style location/speed vectors and the such.

Well, I don't think it would be easy to store the exact values of x,y,z, and vx,vy,vz. This processor would need lots of qubits, I'm affraid, and could be very expensive.

Re:Number Nine

[Junta]

Wow, so you are still running that 386DX/25Mhz? Only way I could see you even attempting to play Return to Castle Wolfenstein is if you had it now. Of course, I would thinnk a 386DX/25Mhz would be incapable of playing RTCW, seeing as how the CPUI is so slow, likely wouldn't support enough memory, and no graphics card it could take would be supported for 3D...

3d hardware was the end ...

[hornet@ch]

I still remember the first 3d hardware, and I still have a pc running with a matrox mystique and a 3dfx voodoo I...

I had been involved in demo coding for a while as an high school student and we had managed to implement a 3d software engine which was working really fine, at higher resolutions as well. The most important thing is, we were having lot of fun.

Maybe I am getting wrong now, but I believe the first version of quake came out without any kind of 3d acceleration (everything was software made, they just wrote an almost perfect code ...)

But one day, well, 3d hardware came out and the whole thing wasn't funny any more. In the beginning it was very difficult for a single person to develop something decent using 3d hardware (because of a lack of good docs), while big companies started to produce lots of games using 3d acceleration, which were very badly optimized.

Well, I don't know, I still think that 3d acceleration took away a big part of the intellectual work due to the optimization process of code in games. Of course there were and there still are exceptions.

Re:3d hardware was the end ...

[Graspee_Leemoor]

The Matrox Mystique was curiously absent from the list of cards in the article.

I remember how much I thought it sucked as soon as I got my voodoo 1 because the Mystique couldn't do texture-smoothing.

graspee

Is this really an event worth tracking?

[MadFarmAnimalz]

Someone mentioned elite, etc. Yes, there was 3D graphics before there were dedicated processors on PCI/AGP cards for this purpose. Going by this ethos, shouldn't we also be celebrating the modularization of the sound support and serial line comms support functions of the modern PCs? Why is the birthday of the 3D card celebrated, and not the ISA/PCI/USB modem, or sound card? Or perhaps Mac users should celebrate the day the monitor was split off the case.

Any processor intensive application will spawn modular add-ons to take some of the burden off the CPU. So long as the task itself, of course, is generic enough to have a sufficiently large market. Basic economics.

By saying there was no proper 3D graphics before the advent of the accelerators, you are doing a great injustice to the demo scene as it was back then. Remember the 256 byte competitions? The 1 kb and 4 kb competitions? Now here were people who knew how to milk code for every iota of juice that was there. The (almost) forgotten art of Code Optimisation.

Heck, there was 3D graphics on my old Commodore 128; I still have Elite. What do you call the original Battlezone? The only difference was, there wasn't any specialised add-on card to do this task on the market back then.

I don't mean to disrespect current makers, researchers, coders, and gamers. I just think there's got to be many more significant birthdays to commemorate.

How about a feature on the demo scene on slashdot? The younger crowd will appreciate the demos, and we'll get these funny comments from the war-torn 386 vets about how they used to make their own transistors out of sand...

Re:Is this really an event worth tracking?

[DeadMeat (TM)]

What do you call the original Battlezone? The only difference was, there wasn't any specialised add-on card to do this task on the market back then.

FWIW, many Atari vector games included extra coprocessors called "mathboxes" for calculating 3D projections. BattleZone was one of the first. Admittedly, it had much more in common with modern FPUs than modern 3D accelerators, but even back then people were dedicating silicon to 3D graphics.

High-End stuff?

[Amazing Quantum Man]

OK, he's talking about mainstream gaming stuff...

But there was some killer high-end stuff for the PC architecture!

Anyone remember the Intergraph workstations? They had custom 3d hardware. In late '98 (or was it early 99?) we had an Intergraph with Wildcat graphics. 16MB framebuffer and 64MB texture (I may have it backwards). Highly accelerated, and killer. We used it to run ballistics and weather simulations.

Re:Number Nine

[crisco]

Yeah that was it, 128 bit memory path to RAMDAC or some such thing. The early '90s, when graphics cards were measured in WinMarks or some such thing, some kind of benchmark on how much they accelerated the basic geometry features. I also remember the arguments over the card drivers that would return done even though the operation was queued, not completed. Gave insanely high 'WinMarks' or whatever, but some people didn't agree with the buffering concept.

I also remember the horrid Windows 3.1 drivers. S3 was known for the best drivers. My Trident card had lousy drivers as well. ATI was notorious for bad drivers, funny how that reputation lingers.

I think they misunderstand Matrox's position...

[Demon-Xanth]

Matrox isn't going away because of the following key features:

1: The best multimonitor around
They starting it, they perfected it, they can do different resolutions under Windows 2000 (they were the first, if not only)

2: Excellent overlay charactaristics
Wanna use a TV tuner card at high resolutions? Ignore nVidia. From my experience programs that run overlay really like Matrox's card, w/ the DVD max feature that allows any overlay to be displayed on the secondary monitor you can port divx video out to the TV. Also, overlay works at much higher resolutions than nVidia solutions have. I don't want to turn my 19" down to 1024x768x16 bit just to watch a DVD, my 14" runs more than that.

3: Acceptable 3D performance, exceptional 3D quality
Although it's not the fastest card on the block, it will still play virtually all games atleast acceptably. And when you are playing them, they have a low amount of artifacts and the textures are well drawn.

4: 2D quality
Although it's much overlooked, it's what most people stare at a majority of the time. Matrox makes thier own boards so they can have a tight control over the filtering components.

I've used a couple S3 cards (low end), Permedia 1 and 2 cards, Riva128ZX, TNT, and TNT2, Matrox MGA, G400 and 450 cards. And so far I have to give props to Matrox for a product that matches my needs. Granted my needs are different from most.
(triple monitors w/ TV tuner and alot of video player programs)

Re:I think they misunderstand Matrox's position...

[RzUpAnmsCwrds]

Bravo. 2D really is better on Matrox cards, isn't it. Got my G200 back in 1999; got my G400 MAX in 2000; G550 in 2001. I have a Radeon for those intensive games but through an innovative solution (switching the monitor cable - hehe) I can still use my Matrox when I'm not playing games. DualHead is, of course, wonderful. I'm thinking of getting a G200-MMS (4 way) - but I'd have to switch my nice 19" CRT for a TFT because of space constraints. Oh well.

Incomplete picture.

[Performer Guy]

There are some serious omissions here.

To ignore the early GLINT work from 3DLabs and not give them their own column in the table is a bit unfair.

The Number9 stuff is missing (no great loss).

Other early work is missing, for example SGIs PC graphics card which predates all of this by about 5 years.

Enough with the Polygons, How about Ray Tracing

[IPFreely]

So all these cards do high speed polygon drawing plus fancy stuff.

Has anyone tried to make a GPU for ray tracing? Good ray tracing scenes can be much better than the scenes drawn by polygon engines.
Yeah, it would mean a whole change of code for current software. D3D would have to change, or maybe have another API beside it, say DirectRay. But the rendering would really get better. Todays hardware should be able to handle the load. And they should scale well also. More GPU's equals more parallel rendering of pixels.

Imagine a truely ray traced virtual world. {shudder with anticipation}

Re:Enough with the Polygons, How about Ray Tracing

[Wolfier]

I fail to see how a ray tracing engine would produce significantly higher quality than todays best triangle rasterizers.

(other than the vast decrease of memory bandwidth usage by a real scene description language than describing stuffs in terms of triangles)

Smoother silhoettes due to the use of NURBS than triangles? Sure. We have T&L - just increase the number of triangles in the scene.

Different lighting effects? Pixel shaders can do that too.

Refraction and Reflection? Even Voodoos could fake that reasonably well with environmental maps.

Both approaches fail to render good-looking global illumination effects without faking it with textures or "ambient" lights.

I say, add support for non-triangle scene description and back that up with hardware-accelerated meshing. Then the quest for the ultimate triangle rasterizer is complete.

The real beginning of 3D hardware

[Animats]

The first real 3D hardware was the Evans and Sutherland Line Drawing System 1, in 1969. This was the first graphics device with a hardware 4x4 matrix multiplier, the basic geometry engine component.

I saw one once, at Case Tech, in 1969. About six racks worth of hardware. Nobody really knew what to do with it.

Re:I remember tons of stuff.

[OpCode42]

Best of all, I remember iSmell.

And thats why you shower regularly!

I thank you...

Best performance

[OpCode42]

My first "3D" graphic card was the Voodoo2. For a while I resisted buying a dedicated 3D card, that was until I saw glQuake ;)

I have tried the ATi Rage 128, Voodoo2/3,nVidia Geforce 2 and nVidia Geforce MX. Without exception, all these cards perform better under Linux than under Windows.

No, this is not a troll. I have got people to try linux after they have seen TacticalOps running on my slackware powered laptop :)

The best example of this was running Return To Castle Wolfenstein on my Geforce 2. It played OK under Win2K (latest DirectX, drivers etc.) - but I had to run it at 800x600 for it to be playable. Running the same binaries under Transgaming's WineX, I could bump it up to 1200x1024 and get a better framerate than under windows at 800x600!

The best supported cards I have owned have been the nVidia cards. Regular driver releases available for both Windows and Linux from their web site - I challenge you to find another gfx card supplier that does the same!

Ok, so part of the drivers are binary only. I say : so what? Nvidia are good at maintaining them, they know the card best and seem happy to support us, so why moan?

Now I'm waiting for my Geforce 4 Ti4600 to arrive... :)

Re:Rendition? Sure I remember...

[Com2Kid]

Yah, GPL was pretty much the ONLY reason that the Rendition card was able to stay around so long.

Pretty darn good game still, hell, damn GREAT game still, heh.

I remember a looking from at a screenshot from GPL printed on the then Highest Resolution Printer In The World(tm)Lexmark z(whatever). Damn nearly looked like a photograph.

When I saw screenshots of it directly, hell, it DOES look damn nearly like a photograph!

Whatever API they used and however it interacted with that chip was damn powerful, blew the living shit out of anything to come for another 2 years or so.

More than 7 years ago.

[pinkpineapple]

Wasn't Mistery house on the Apple ][ the first 3D game of all time? It was published around 1980, I believe.

PPA, the girl next door.

Matrox could really clean house....

[MtViewGuy]

As a current user of a Matrox G400 DualHead AGP 32 MB card, I can definitely say that the 2-D graphics quality of this card--like all Matrox cards--is second to none. Not even the latest ATI Radeon 8500 series comes close to the amazing sharpness of 2D graphics that the Matrox cards now offer. I've seen the output of the better quality cards that use the nVidia chipsets and they (on the average) don't even come close to the crisp display quality you get from Matrox cards.

This is why I'm REALLY hoping that Matrox does make another stab at a high-end 3-D graphics card that can compete against the Radeon 8500 and GeForce4 Ti4200/Ti4600 series but still offer the unrivalled 2-D display quality Matrox is famous for. Using the modern 0.13 micron process to make the next-generation Matrox chip, they could easily offer industry-leading graphics acceleration and MPEG-2 decoding equal to that of the GeForce4 series. Such a card--even if it costs slightly more than the cards that use the GeForce4 Ti chipsets--would be instantly lapped up by gamers who want the clearest graphics display.

Re:Hmmmm Matrox looks pretty good...

[MtViewGuy]

I agree with your assessment of Matrox cards.

I use a Matrox G400 DualHead AGP myself and frankly, I have YET to see a graphics card using the ATI or nVidia chipsets that match the amazing sharpness of the current Matrox AGP cards on 2-D graphics. These cards are perfect for business users, desktop publishers and CAD/CAM users, where picture quality takes precedent over 3-D graphics speed.

This is why I really want to plead to Matrox to develop a no-holds-barred 3-D graphics card that can match the ATI Radeon 8500 card and any card that uses the nVidia GeForce4 Ti4400/Ti4600 in terms of 3-D graphics acceleration and also assist in MPEG-2 decoding for DVD movies, but still maintain the legendary display quality sharpness Matrox is famous for. I can say that even if the resulting card costs slightly more than cards that use the GeForce4 Ti chipsets you know gamers are going to lap this potential Matrox card up in a New York minute, mostly because many gamers have been disappointed by the sub-par sharpness of ATI and nVidia chipset graphics cards.

Re:Number Nine

[Some Dumbass...]

You only need 16MB to handle the highest resolution computer graphics displays ever made.

This is true for 2D displays, but when you start having double and triple buffering plus z-buffers it starts to add up. Then add the texture requirement and you can see why most newer cards have 64-128MB of memory on the cards.

Technically, though, he's still correct. Sure, your card would be awfully slow playing something like Quake III without on-card memory for textures, etc. :) However, any 3d rendering could theoretically be done in software and then (slowly) displayed on a 2d-only card, right?

Re:I'm a 3D gaming junkie....

[Wolfier]

I've accustomed to buying stuffs 1/2 to 1 a generation behind the technology and so far it saves me quite a sum and apparently I'm not missing much at all, since (surprise) games usually lag behind hardware advances.

Also, older cards can be sold off the usenet when you upgrade. Did that for quite a few times (I sold my stuffs at half the price I was going to pay for my upgrade, and it worked) until one day I decided to put all in-use hardware into my girlfriend's computer when I upgrade, then sell hers. This chain has been working pretty well.

Re:Video Card Woes

[Wolfier]

well if you're only using Outlook and Word or Pine ;) then no.

However, you can really open your eyes by paying $10 for a used i740 card.

Re:Ah, 3d hardware. The beginning of ugly.

[Tofuhead]

As an avid console gamer who had stopped gaming for a couple of years after the PSX succeeded the throne of console dominance from the Super NES, I understand _exactly_ what you're talking about. In my absence from the gaming world, I lamented the "death" of 2-D at the hands of ugly, boring, primitive 3-D graphics. Even within the SNES era, people raved over games like StarFox, a 3-D game which had very little appeal for me, but which for many was a vision of how games should look and play. Meanwhile, I foresaw that it would be many years before 3-D graphics would even approach the beauty of sprite-based graphics, and that's turned out to be true IMO.

However, it should be noted that there are examples of games that utilize 3-D graphics while maintaining 2-D gameplay and feel to great effect. One example that comes to mind for no real reason is ThunderForce V, which is a great horizontal-scrolling shooter (aka shoot-em-up or "shmup") that uses 3-D graphics which are small enough to be somewhat detailed. The kicker comes when encountering boss enemies, where the camera seamlessly zooms and rotates around the scenery from the standard side view, taking obvious advantage of the 3-D nature of the graphics.

I now happen to enjoy a lot of games that have made the switch, in all sorts of genres. Some quick examples include Mario (platformer), Zelda (action RPG/platformer), Final Fantasy (RPG), Hundred Swords (SRPG), etc. Street Fighter EX in any incarnation will never be able to replace its 2-D progenitor for me, but in many other ways, I've come to tolerate 3-D graphics in games where its usage adds more to the gameplay than it detracts from the visual appeal.

In this last regard, I think Nintendo's Legend of Zelda: Ocarina of Time was revolutionary for me (as it's the one game that brought me back to console gaming). It was full of rough graphical edges, but to be fair, its immediate 2-D predecessor on the Super NES used graphics that were small and fairly undetailed, and in comparison were less impressive overall at the time. Ocarina of Time is still beautiful, and 3-D graphics helped that game achieve incredible depth, as well as a fantastic sense of the sheer vastness of the game world.

< tofuhead >