Slashdot Mirror

Permedia and GLint were chips from 3Dlabs. They were a british company, I remember working with these chips around 1996, so it's been a while. They are not in the graphics cards business anymore:

"On February 24th 2006, 3DLABS refocused its business and stopped developing its workstation graphics cards for the PC and announced it would focus on its new DMS(TM) low-power media-rich application processors." ... http://www.3dlabs.com/content/Legacy/

The problem with the PXA versions of the ARM based SOC's has been proprietary and/or binary only libraries. Linux runs on the core cpu but you run into a wall when you need to get multimedia codecs or 2d/3d graphics support working at any usable rate.

RMI Mips based (formerly AMD/Alechemy) SOC's http://www.razamicroelectronics.com/products_alchemy/ are more open when it comes to multimedia and Linux support.

3DLabs has some multicore ARM mutimedia 2D/3D SOC's http://www.3dlabs.com/content/mediaProcessor.asp . But they don't open the tools and libraries to develop codecs.

Freescale also has their i.MX line of ARM media SOC's http://www.freescale.com/webapp/sps/site/taxonomy.jsp?nodeId=0162468rH311432973ZrDR

Debian/Linux and UBoot support is available for the cores for many ARM SOC's but the problem has always been open source with the multimedia and graphics acceleration portions of the designs.

Tssss.....

http://workstation.3dlabs.com/products/product.asp ?prod=293

Well you can check the link to be sure -- the only thing that struck me about that IBM box is that it has a pretty nice graphics card, although I don't think it's the NVidia one that you're thinking it is.

It's the "3Dlabs Wildcat Realizm 800" which is apparently a dual-GPU, PCI Express monster of a card (640MB RAM) that takes at least 2, if not 3, card slots because of its thickness [2], and is designed to either drive two monitors independently or drive a stereoscopic system. Still, it only lists separately for about $2k, so it doesn't go that far in justifying the Intellistation's price tag. Plus at least according to Anandtech [1], it's outperformed by the NVidia Quadro 4400. So it's not even the ultimate top of the line (according to them). The only really special thing I've seen about it is that it does video genlock, so you can use an external "house clock" source if you were working with digital video. [3] However that doesn't really mesh with the application for a big IBM workstation, at least that I've ever seen. But what do I know.

If anyone wants to offer a good explanation of why the IBM Opteron workstations cost so much, except for the three letters on the front, I'm curious. But I sure as hell can't figure it out.

[1] http://www.anandtech.com/video/showdoc.aspx?i=2383 &p=4
[2] http://www.computerarts.co.uk/news/3dlabs_unleashe s_wildcat_realizm_800
[3] http://www.3dlabs.com/products/product.asp?prod=29 3

Hopefully not this.

-:sigma.SB

If your currently using SGI, then I would assume that your not looking for a gaming card. If your doing 3D design work, then you might look into The Wildcat Multiview Card. It only works in conjunction with the Wildcat Realizm cards. Not cheap, at $825 for a midrange card, and nearly $3,000 for the top-of-the-line card and another $420 for the multiview card, but then they are professional cards, not intended for hobbiest or gaming machines.

If your currently using SGI, then I would assume that your not looking for a gaming card. If your doing 3D design work, then you might look into The Wildcat Multiview Card. It only works in conjunction with the Wildcat Realizm cards. Not cheap, at $825 for a midrange card, and nearly $3,000 for the top-of-the-line card and another $420 for the multiview card, but then they are professional cards, not intended for hobbiest or gaming machines.

Nope, 512 mb can sound 'too much" for gamers but, current trend is 768 mb in pro market.

http://www.3dlabs.com/products/product.asp?prod=29 3&page=6

For example. Speaking about overkill, you can use the gfx card RAM on some systems when its free. Its of course, PCI Express amazing bandwidth does it.

Funny is, to check FPS of some games on such cards to prove further FPS is a lie like Mhz. Years ago, we tested Quake III on such a system and the FPS levels were below the latest home gaming card :) Of course, you feel like bowing to Carmack when you see couple of that freak cards features being used.

Our GFX Artist sure did :)

To complement your nice Tyan motherboard, get one or two (XXX check for physical sizes) of them realizm 800 from 3dlabs. They are the only 16 lanes PCIexpress videocards I know of. Not sure about GPGPU, but at 3840 x 2400, solitaire is bound to look amazing... especially if you can get some nice 9.2Mpixel displays as well: High end videocard without a matching display, what would be the point? Check for instance the IBM T221).

Anandtech reviewed the Realizm 800 here.

Mhhh... If you wanted the machine to be a server of some sorts, then I just wasted 10 minutes typing all this!

Im suprised people dont mention 3dlabs line of cards. I got a vx1 for a friend back in 98 for his CAD box and he swears that it still rips geforces and radeons latest offerings in opengl for autocad.

link here

Here's an example of a card with more than 512MB's of RAM, the "Wildcat Realizm 800".

And the majority of you dare to call yourselves "Uber-geeks" while drooling over commodity hardware, pah!.

And what's more, the Wildcat Realizm 800 has 640MB of memory. Anyone can buy it, but it costs around $2,000.

ATI is pushing the texture barrier by incorporating 512MB in their newest X850 video card lineup.

Because, as we all know, video cards never have had more than 256MB of memory before...

3Dlabs Realizm 800 with 640MB memory.

Exactly, but the Wildcat Realizm 800 has the 640meg version as well.

Nothing new...

640MB GDDR3 total memory

512MB GDDR3 unied memory with 512-bit-wide interface bus

128 MB GDDR3 DirectBurst memory with 128-bit-wide interface bus

Full Specs Here

It's the 3D-Labs Wildcat Realizm 800, and it's PCI-Express too.

NVIDIA's just announced GeForce 6200 cards with TurboCache - the first graphics cards that truely take advantage of the PCI Express bus by using system RAM to store textures."

BZZT, WRONG. Here is the first PCI Express video card that stores textures in system memory.

(For that matter, 3Dlabs were the first to release an _AGP_ card that stored textures in system memory: anyone remember the Oxygen chip?)

Unless, of course, you mean to tell me this is the card I should be replacing my 9800 AIW with.

With regards to driver support, where is 3dlabs' support of OS X anyways, I mean, it's easy to get Linux drivers for high end cards. Googling for support for the same company for OS X was much less fruitful. Additionally, licensing has everything to do with drivers. Were nVidia and ATi much more open with the distribution of their binary modules, it would be easier for desktop-oriented distro makers to create a better, more consistent product. Apple can ship with ATi drivers in the box; there's no need to go to ATi's site to download drivers; ATi, and nvidia, don't afford the same luxury to Mandrake users.

Try using a real desktop-oriented distro some time, you'll see that your experience of Linux on the desktop has been soured by trying to use an environment designed for servers to perform desktop tasks. Other distros are a lot more friendly.

Actually, high-end is closer to $3,000. If you didn't know that, it's obviously because you're a lower-class monkey, who thinks his $70,000/year salary is special.

Well, he could be buying one of these, which is about the same as a payment on a $600,000, 30 year mortgage, with 6% interest, and 20% down.

OpenGL 1.5 with full arb extensions is basicly OpenGl 2.0. At least you can program the OpenGL 2.0 way.

Take a look at the Shading Language Demo

You will see what I mean

"The Wildcat Realizm 200 boasts a VPU and 512 MB of onboard memory, the largest graphics memory available on an AGP 8x-based graphics accelerator to effortlessly handle large models and textures. Wildcat Realizm 200 is equipped with two, dual-link DVI-I connectors, which are ideal for driving multiple high-resolution professional displays from a single graphics accelerator, including the dual high-end 9.2 Megapixel displays."

Of course, 2 of these cards in Alienware's new dual configuration PCIe motherboard would probably produce quite the gaming setup. You might even be able to spend 10 grand on your gaming system! Think how cool you would be then!

Ring ring!
2 GPUs on one card from 3Dlabs:
Wildcat Realizm 800

That's exactly what 3Dlabs' Wildcat Realizm 800 is. 2 GPUs, lots of memory, in a single slot.

3Dlabs did it too with Wildcat 4 and the upcoming Realizm VPU/VSU.

GLSL is clearly OpenGL 2.0's killer feature. Check out the GLSL section of 3Dlab's site for some resources. The slide show from Randi Rost's book tour are on-line as a 9 MB PDF. Before too long, you should be able to download RenderMonkey, a cool tool for playing with GLSL and D3D shaders.

Unfortunately, NVIDIA's GLSL implementation has a ways to go. You won't even get the necessary ARB extensions unless you set a debug registry flag (google "ShaderObjects"). After that, good luck.

I'm anxious to see how the industry receives GLSL. Will we get some decent drivers before Doom 4?

On the subject of other graphics card contenders, what about 3D-Labs, home of some of the world's most powerful graphics cards? Something tells me that 3D-Labs' Wildcat4 7210 offering 4 monitor outputs, 384 MB DDR RAM, and allegedely decent linux support is something worthy of notation. While the average price of a 3D-Labs card isn't consumer level, they are still in competition with ATI, nVidia, et al. in the field of professional quality graphics cards.

I don't know if it was the actual cause (certainly might have been) but buying them seemed to distract the company for long enough for NVidia to take over.

My first accelerated gaming card was a Quantum 3D Obsidian 100SB which was a funny sort of Voodoo1.5 pass-through card (never saw another one). I also had a horrible Virge3D card which was slow for everything, STB card, Cirrus Logic, Trident, you name it. I still run a Voodoo3 in m y 200MMX system but haven't played a game on that one in AGES...

Anyone remember the 3DLabs Permedia cards? They never worked right for games (had missing geometry and texture for unsupported nodes) but were one of the first and fastest serious accelerated OpenGL boards. They still appear to be in business but they are still professionally oriented, not game oriented.

Yeah, printf works real great for coding in OpenGL 2.0 Shading Language.
</sarcasm>

They should have also benchmarked the latest 3Dlabs cards in order to give us a proper frame of reference. For all we know, both these cards could be providing inferior performance compared to the latest Wildcat; good gaming performance doesn't necessarily translate into a good professional video card.

The Wildcats are also cheaper: $899 for the 512 MB VP990 Pro and $499 for the 256 MB VP880 Pro or the 128 MB VP970 (from the 3Dlabs eStore) compared to $530 for the cheapest 128 MB ATi FireGL X1 and $1250 for the cheapest 128 MB nVidia Quadro FX 2000 (the 256 MB variant was used for benchmarking).

Anyways, these aren't even ATi's and nVidia's top of the line cards; ATi's is the FireGL X2-256 and nVidia's is the Quadro FX 3000.

They should have also benchmarked the latest 3Dlabs cards in order to give us a proper frame of reference. For all we know, both these cards could be providing inferior performance compared to the latest Wildcat; good gaming performance doesn't necessarily translate into a good professional video card.

The Wildcats are also cheaper: $899 for the 512 MB VP990 Pro and $499 for the 256 MB VP880 Pro or the 128 MB VP970 (from the 3Dlabs eStore) compared to $530 for the cheapest 128 MB ATi FireGL X1 and $1250 for the cheapest 128 MB nVidia Quadro FX 2000 (the 256 MB variant was used for benchmarking).

Anyways, these aren't even ATi's and nVidia's top of the line cards; ATi's is the FireGL X2-256 and nVidia's is the Quadro FX 3000.

They should have also benchmarked the latest 3Dlabs cards in order to give us a proper frame of reference. For all we know, both these cards could be providing inferior performance compared to the latest Wildcat; good gaming performance doesn't necessarily translate into a good professional video card.

The Wildcats are also cheaper: $899 for the 512 MB VP990 Pro and $499 for the 256 MB VP880 Pro or the 128 MB VP970 (from the 3Dlabs eStore) compared to $530 for the cheapest 128 MB ATi FireGL X1 and $1250 for the cheapest 128 MB nVidia Quadro FX 2000 (the 256 MB variant was used for benchmarking).

Anyways, these aren't even ATi's and nVidia's top of the line cards; ATi's is the FireGL X2-256 and nVidia's is the Quadro FX 3000.

It is just me or did the soundtrack seem, well, japanese? The movie in general is just flat-out weird. Not quite funny. Just weird.

I'm not a big fan of Halo myself. I've never played it. None of my friends have. There's just not that sense of greatness about the game that makes me feel like I really need to give a crap about it. But something is worth noting about the game; and that's the sudden deluge of movies circulating the internet that were recorded using the game. This is probably due in part to the game's amazing graphical capability, of which the X-Box itself is very likely playing a key role in, but perhaps also due in part to something easily facilitating video recording? I read something awhile back (probably one of the OpenGL 2.0 white papers about pixel pack/unpack from 3D Labs, maybe not, my brain's starting to fall asleep right now) that seemed to underline the fact that you can send a lot of data to a video card and have it be all accelerated and fast and all, but it's not quite as easy to send it back into memory for, say, recording high-quality 3D rendered movies in real-time. That all we'd need is some kind of bus or compression scheme to get the stuff back into main memory from the video card and all of a sudden you'd have this massive flood of 3D-rendered indie movies.

That said - it is quite likely that this, and the others like it (if you haven't already, check out Red vs. Blue and see what I'm talking about here), is the tip of a very big iceberg.

i believe there are some sites that use gaming benchmarks to review these cards, try highend3d.com, they usually are a good place to start looking into it. if not, google is your friend.

3Dlabs came out with its Wildcat VP graphics cards a while ago. What makes these unique is their "Visual Processing Architecture" (formerly "P10"). It is composed of hundreds of small, relatively simple SIMD vector and scalar units, allowing for extreme scalability and re-programmability.

New features that would require hardware revisions with traditional graphics architectures only need a software update with the Wildcat VPs (3dlabs gave the example of Matrox's ingenious fragment anti-aliasing, which could be added in this way, though this has not been done to my knowledge). In addition, the less-specialized nature of the hardware also allows for the acceleration of things such as Photoshop filters (Rampage anyone?). The use of VPU's allows for rapid software modification of the hardware.

The latest model, the Wildcat VP990 (not out yet) doesn't compare to the Geforce FX 5900 Ultra in terms of DX 9 compatibility (it's not a gaming card, it's for CAD/CAM and other high-end graphics applications) and pure polygon thoroughputs, but its potential for flexibility is unmatched.

Had the original poster READ the article he was posting about rather than skimmed, he would have seen that the test pitted top of the line consumer grade X-86 hardware against top of the line Dual G4 Mac hardware... to quote the article:

"The showdown pitted a single-processor Dell 3.06GHz Pentium 4 and a 1.25GHz dual-processor Power Mac G4 (the fastest Mac then available). The contest compared renderings of files created in Adobe After Effects, Illustrator and Photoshop software." Source:

Incidentally, a single processor P4 machine is not by any means top of the line PC hardware. Perhaps they are too embarrassed to show what would happen with top of the line X-86 hardware (non-server class, lets stick to workstation vs. workstation hardware)... A Dual Intel Xeon processor (Xeons are currently running at a peak of 3.06 Ghz) box with a workstation class graphics accelerator instead of the gaming graphics cards you can get for the mac (GeForce 4 IS a gaming graphics card!... THIS is a workstation graphics card)... I betcha price is comparable at that level of X-86 hardware.

Not to mention, to add insult to injury for Apple, the single processor Pentium 4 3.06 Ghz PC (which I'm sure retailed for $1000 - $2000 USD less than the Apple box) whipped the Power Mac in EVERY category of the comparison. I'm sure Apple's own proprietary "equivalent" software runs faster on MacOS than the Adobe software. Perhaps they should open some of the tricks they are hiding to accomplish that to Adobe, one of the companies that made Apple what it is today!

Oh yeah, and one day, if Apple has the balls, they should compare top PC-Workstation hardware to top Mac-Workstation hardware. To make it fair (and cut the whining), limit the PC-Workstation to the retail price of a top of the line Mac-Workstation (that is currently $3,799 without a monitor!)... then compare those machines and see the embarrasing truth (well, embarrasing to Apple, who claims that the "turbocharged Power Mac rips through digital video and 3D projects faster than Pentiums can say 'uncle.' " Source:)

I wish Apple would move to X86... if they can convince people of that much BS and stir up what can only be called religious Mac worship, they would probably do great (and make much more profit) if they switch OSX to X86 and built the same PowerMacs on X86 hardware... maybe the Opteron, who knows :) They've got enough of a name where they could just pull it off and, aesthetically speaking, they are ahead of the X86 world, for now. Plus, being *nix based now, it should be a simpler move than it would have been before!

I think I'd rather have a 3dLabs Wildcat4 7210, thanks.

Having an overclocked gamerz card is a bit less impressive than a professional one - a bit like an x86 weenie with an overclocked celery comparing his computer to an Alpha SMP box...

That's utter Bull. Pure and simple. Often I'm astonished at what rubbish a supposed geek site like /. posts on stuff that should lie within it's 'area of expertise'.
Aside from the fact that this piece is offered by somebody infamous for pushing the envelope in crappy hardware, I seriously doubt that it beats all-time, all-star leading edge GFX hardware like, for instance, the FireGL 4 or the newest Wildcat.
Gawd, I hate these n00by statements...

It's huge (eats a PCI-slot) and noisy

Many workstation-class video cards take up multiple slots:

www.sun.com/desktop/products/graphics/xvr1000/
www.3dlabs.com/product/wildcat4/index.htm
www.3dlabs.com/product/wildcat3/wc3_6210.htm
mirror.ati.com/products/workstation/fireglx1/index .html

I have to wonder, though, how many free PCI slots you have in your case. Do you have one? Two? Three? All of them? With most components integrated, I would expect the average user to have three or more PCI slots free. Yes, some will buy FlexATX or MicroATX boards, but then they're buying for a specific purpose and even then I think it's unlikely they'll use all the slots.

It's huge (eats a PCI-slot) and noisy

Many workstation-class video cards take up multiple slots:

www.sun.com/desktop/products/graphics/xvr1000/
www.3dlabs.com/product/wildcat4/index.htm
www.3dlabs.com/product/wildcat3/wc3_6210.htm
mirror.ati.com/products/workstation/fireglx1/index .html

I have to wonder, though, how many free PCI slots you have in your case. Do you have one? Two? Three? All of them? With most components integrated, I would expect the average user to have three or more PCI slots free. Yes, some will buy FlexATX or MicroATX boards, but then they're buying for a specific purpose and even then I think it's unlikely they'll use all the slots.

I am not trying to be a troll here but what exactly do you mean by top-of-the-line cards? Here in America we pay over $3000 for top-of-the-line cards

3dLabs WildCat II

I think what Richards is overlooking in his commentary is that Cg is not *supposed* to be a general-purpose graphics programming language. Its design goal was precisely what he said later in the article -- to expose the capabilities of current (and presumably future) NVIDIA hardware without requiring programmers to write assembly code. Likewise, conditionals like if, case, and switch aren't in there right now because the profiles the compiler is aimed at -- DirectX and OpenGL extensions -- don't yet support them. I expect this to change.

Also, Cg programs run at the level of vertices and pixels. This is the wrong place to be thinking about a scene graph: that happens at a much higher level of abstraction. Dealing with scene graphs in a fragment shader is a little bit like making L1 cache aware of the memory-management policy of whatever OS happens to be running.

After reading the article a few times, I think it's meant more as a "here's why our product is better than theirs" release than an honest criticism of the design of Cg. If he was interested in the latter, there are a few obvious issues. I won't go into them all, but here are two I ran into last week at a Cg workshop:

• Cg limits shaders to single-pass rendering. This is a design limitation: there are lots of interesting multipass effects, and it's not all that difficult to get the compiler to virtualize the shader to do multipass on its own. The Stanford Real-Time Shading Project people wrote a compiler that does precisely that: it uses Cg as an intermediate language. The advantage of that design decision is that you-the-programmer have fuller control over what's happening on the hardware, which is the entire point of the exercise.
• Cg requires that you write separate vertex and fragment shaders. You can't do things like texture lookups inside a vertex shader; you can't change the position of your vertices inside a fragment shader. Again, this gives you control over the details of the pipeline at the cost of some added complexity. This can be changed by changing the semantics of the language.

One final note: Cg is not the be-all and end-all of real-time shading languages. Nor is DirectX 8.1, 9, or whatever. Nor is the SGI shading language. Real-time shading on commodity hardware is still a new enough field that the languages and the hardware are evolving. DirectX 9 and OpenGL 2.0 both incorporate shading languages that will by nature be less tightly coupled to one vendor's hardware. Watch those over the next year or so.

This guys got some good points, espically with regard to pipelines with more progammibility such as the PlayStation 2 gpu. 3Dlabs is also developing highly progammible pipelines
3DLabs press release

Having better control structures and pointers will be important down the line.

This has been around for a while on Wildcat cards - you know, the ones that cost many thousands of dollars. If you're talking about the "industry", though, you can't leave them out.

Here's a URL:

Is there a layman's reference I can read about how OpenGL 2.0 works?

There's a very nice overview with more information here.

(google is your friend.)

Is there a layman's reference I can read about how OpenGL 2.0 works?

There's a very nice overview with more information here.

(google is your friend.)

All of this leaves me a little bit confused. I'm not sure why we need two (or, perhaps, more) C-based shader languages, at least one of which (Cg) is hardware-specific, but API neutral.