Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Asus Crams Three GPUs onto a Single Graphics Card

Posted by Zonk on Thu Apr 10, 2008 01:43 PM
from the when-isn't-cramming-a-good-thing dept.
Barence writes "PC Pro has up a look at Asus' concept triple-GPU graphics card. It's a tech demo, so it's not going to see release at any point in the future, but it's an interesting look at how far manufacturers can push technology, as well as just how inefficient multi-GPU graphics cards currently are. 'Asus has spaced [the GPUs] out, placing one on the top of the card and two on the underside. This creates its own problem, though: attaching heatsinks and fans to both sides of the card would prevent it from fitting into some case arrangements, and defeat access to neighbouring expansion slots. So instead, Asus has used a low-profile heat-pipe system that channels the heat to a heatsink at the back of the card, from where it's dissipated by externally-powered fluid cooling pipes.'"
+ -
story

Related Stories

This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • Drivers first. (Score:5, Interesting)

    by San-LC (1104027) on Thursday April 10 2008, @01:48PM (#23028016)
    The technology for multi-GPU processing is already out there (SLI, Crossfire), and now the companies are trying to increase the number of GPUs that can be daisy-chained (CrossfireX, 3-way SLI).

    However, it seems with all of these methods, the weak link is always driver support. I think that drivers will have to develop further before anything like this can take true form and be useful.

    As an aside, did anyone notice that half of the Slashdot description sounded like an advertisement for Asus GPU cooling?
    • Re: (Score:2, Insightful)

      I an content with the drivers that I have. However, I am still looking for a decent high end card that does not need two slots in my case. How about fixing heat and size issues first?
      • Re: (Score:2, Insightful)

        I am still looking for a decent high end card that does not need two slots in my case. How about fixing heat and size issues first?

        Bingo. I thought the main point of multi-GPU graphics cards (and multi-core processors) was to build good gaming rigs (and workstations) without having to use a monstrous extended ATX uber-tower with multiple CPU sockets and video card slots.

        Improved manufacturing processes and software/drivers have allowed us to have multiple processor cores and GPUs in a shoebox-sized Shuttle XPC. Asus's big, hot, and inefficient card just shows us that current manufacturing processes and software/drivers aren't ready

    • Re:Drivers first. (Score:5, Insightful)

      by eebra82 (907996) on Thursday April 10 2008, @02:00PM (#23028158) Homepage

      As an aside, did anyone notice that half of the Slashdot description sounded like an advertisement for Asus GPU cooling?
      Advertisement or announcement? Does it really matter, since most news items could be considered as advertisements?

      I think the talk about the cooling is important since one of the most difficult tasks is not how to get three GPUs on a single chip, but to get a viable cooling solution that doesn't sound like a vacuum cleaner and one that doesn't require too much space (or it would essentially kill the whole concept).
      • Re:Drivers first. (Score:4, Insightful)

        by Applekid (993327) on Thursday April 10 2008, @02:33PM (#23028532)
        Unfortunately Nvidia set the pace for all video card manufacturers when they spouted they were going to do 6 month product cycles. Trying to go faster than Moore's Law had resulted in top end cards being massively more expensive than yesteryear's top end cards and taking exponential amounts more of energy than they used to... to the point where the high end can't even sustain itself without not one but TWO specificly designed supplmental power leads from the PSU.

        Multi GPU is the only way to keep that breakneck pace, just like the CPU world is trying to deal with hitting the wall (or, depending on who you ask, the low hanging fruit has already been picked). But the penalties for the reach exceeding the grasp is absolutely catching up with them.
      • Re:Drivers first. (Score:5, Insightful)

        by timeOday (582209) on Thursday April 10 2008, @03:52PM (#23029554)

        Advertisement or announcement?
        It's neither. In fact isn't nothing more than a report of a failed experiment - I quote, "A standard 512MB HD 3850 running our Crysis benchmark in high detail at 1,280 x 1,024 averaged 26fps, while switching to the X3 increased that score by just 3fps."

        In other words, it doesn't work! I'll worry about cooling 3 GPUs when they are at least able to do something useful! Until then I would cool this board by unplugging 2 of the GPUs and enjoying practically the same performance.

    • Re: (Score:3, Interesting)

      the weak link is always driver support

      Kinda sorta. Splitting rendering across multiple GPUS has afaict become much harder lately. GPUs used to be mostly fixed function pipelines, while the current generation has more in common with programmable stream processors (e.g., shader programs).
  • onionistic (Score:5, Funny)

    by Anonymous Coward on Thursday April 10 2008, @01:54PM (#23028092)
    Fuck Everything, Were doing five cores.
  • So does that mean we can play Crysis now?
  • by Forge (2456) <forge@myrealbox. c o m> on Thursday April 10 2008, @01:58PM (#23028124) Homepage Journal
    Remember when All razors had a single blade? Then double blade razors were all the rage. These days, Triple and quad blade razors are around. Soon we will have 5 blades but I would call that a cheese grater.

    Same thing with CPUs and now GPUs. Problem is, at what point dose it become a pissing contest rather than a way to provide more performance for an application that needs it.

    And speaking of those demanding applications. Am I the only one who notice that some of the latest video games running on the best available hardware provide no improvement in appearance or game-play over older games of a similar type running on older hardware?

    It's bad enough that I am tempted to think the programmers are just adding fat to make sure the game demands a more expensive video card.

    Kevin.
    • Re: (Score:2, Informative)

      But...we have a razor with six blades [gillettefusion.com] already.
    • In defense of the razors, I absolutely love my Fusion (4 blades). It's each time they add a blade they reduce the spacing between the blades by a similar amount. So each blade only has to take off 1/4th of what a 1 blade model would.

      In a BadAnalogyGuy way I do hope that some computers (especially laptops) move in this direction. Why do I need a 2 gHz dual core processor for my EEE style laptop. Break it into a cheap, slower, power efficient general processor then have a few other small, cheap, power efficie
      • by Hatta (162192) on Thursday April 10 2008, @02:28PM (#23028476) Journal
        I don't buy that. The first razor is always going to be doing most of the cutting because it's in front. The others are just there for show.
        • by 0100010001010011 (652467) on Thursday April 10 2008, @02:46PM (#23028704)
          Say what you want. I used to always get the cheap $1 razors when I was in college because I was cheap. But the Mach 3 and furthermore the Fusion make a great improvement. Even when I shave every other day it still goes through it smooth and with no 'tearing' on my face.
          • I think we've had this before, but let me add another vote for the Mach3.
          • by Cheeko (165493) on Thursday April 10 2008, @03:05PM (#23028974) Homepage Journal
            That probably has more to do with the quality of the individual blades. Cheap $1 razor will cut like a cheap $1 razor.

            That being said I used to use the 2 blade Gilette razor and have since moved on to a 3. What I have noticed is that it does the job faster and the overall blade lasts longer. What i suspect is happening is that the first blade may dull, but its making the rough cut anyway, then one of the other blades which is sharper follows up with a cleaner cut.

            I think the more important advancement has been all the other stuff on the blade head. the mounted springs, the lube strip, the rubber precut strips that tension the skin, etc. I suspect all those contribute more to the newer blades being better.
        • Re: (Score:2, Interesting)

          Thats true... "to a point"... but when that first point wears out, and starts just pushing the hair over, the second blade snags on, cuts the hair on an angle, the third cuts it almost straight (flush) and the forth cuts it flush...

          Also, how you roll the blade has some effect on which blade gets the first cut... similar to a surf board, its flat... however when you put your weight to the back, its only the back of the board thats touching the water...
    • by jellomizer (103300) on Thursday April 10 2008, @02:12PM (#23028296)
      Well to be fair the new games do have better graphics. But the problem comes down to the fact that graphics are improving beyond the average persons eye and interpret graphics. Much like sound cards a decade ago. We had some major improvements from the Ad-Lib up to the SoundBlaster 16. But after that even though the cards have massive improvements average joe doesn't know the difference. The same thing is happening to video cards now. The new games that fully utilize the card adds effects that are very subtile, or on the old systems they cheated to add the effects (Fixed background Bitmaped Images, A semi-transparancy layer to simulate haze. While now the background is getting nearly fully rendered so if there is a mountain in the background and you have a powerful enough graphics card and a high enough resolution you can see each blade of grass on that mountain, and the Haze is like real haze not as uniformed as before more like real life. But the average Game player wouldn't really focus on these details, if they are actually interested in playing the games.

      Unlike the old days you can see a huge difference between a CGA, EGA, VGA and Super VGA. Then Super VGA held on for a while then the 3d Cards started coming out and there were huge improvements even now. But I think we are getting to a point again where the details they can produce is beyond what is needed.
      • But I think we are getting to a point again where the details they can produce is beyond what is needed.

        Which is probably why we're getting a lot more chatter on the raytracing issue. I believe that'll be the next big step.
        • by jellomizer (103300) on Thursday April 10 2008, @02:44PM (#23028680)
          While I am not sure why I got a Flamebate mod on my post. But Raytracing real time could have advantages on cleaning up issues that we do see. While now we can see each individual grass blade each grass blade looks and moves like a broken stick. But will it be good enough for a competive advantage. Or is it that the graphics artests are not good enough for realistice 3d images. Even the stuff that takes months to render for the movies still looks computer generated and seem unrealistic when things start moving. Much like the Final Fantasy movie (yea it is old but it is a great point) they showed adds before it was released with the people face as a solid image, and you couldn't tell if it was real or not. But when you see them moving and talking you knew it was fake, and lifeless. Pixar worked around these issues by not making people look realistic they use a cartoon aproach to make them seem like cartoon and have them different enough for us to connect with. But it is an issue is it performance and CPU or the fact that we don't have artests good enough for the work yet.
          • While I am not sure why I got a Flamebate mod on my post

            You probably pissed off somebody with mod points and they're taking it out on you, regardless of the post's contents. It happens. Maybe I'll catch it in M2.
      • by default luser (529332) on Thursday April 10 2008, @02:37PM (#23028592) Journal
        I have to agree, I'm seeing less and less use for a higher-resolution screen for home/gaming use.

        For games on the desktop, the maximum resolution you have to push (realistically) is 1920x1200 (really, anything larger at 2-3 feet away is overkill), and the maximum resolution you have to push on a television (if you're into that) is 1920x1080. Funny, midrange $150-200 cards can do that today, with high quality, in all games except Crysis.

        So yeah, I can see the slowdown in graphics tech coming around. The fact that you can play any modern game in medium settings at 1280x1024 with a $75 add-in card shows us exactly why we're hitting the developmental wall. Most people are happy with our current level of graphics, and the cost of new graphics architectures rises exponentially with every new revision; so, if you don't have the demand, you're not going to rush production on the next-generation of GPU architectures.

        Unfortunately, this leaves the %1 of hardcore gamers bitching, and they tend to bitch the loudest, so Nvidia and ATI are trying to placate them with stop-gap SLI solutions.
        • Well Hard core gamers won't be happy until they have something like the holodeck from ST:TNG, without the random failures to actually make what your doing a real problem vs. having fun.
      • His point is perfectly valid. I don't get nearly as excited about recent jumps in graphics technology because the difference in quailty is not as stunning as you'd once get.

        The difference between CGA and my Amiga was immense. The difference between no antiscopic filtering and 16x antiscoping filtering is best left to those with 20/10 vision.

        Once the pixels got indistinguishably small, and the hues varied to the limit of human perception, we were left only with increasing art quality, animation, lighting and
      • by vux984 (928602) on Thursday April 10 2008, @04:18PM (#23029804)
        Well to be fair the new games do have better graphics. But the problem comes down to the fact that graphics are improving beyond the average persons eye and interpret graphics. Much like sound cards a decade ago.

        I think the biggest irony is that in multiplayer competitive people disable all these features anyway because

        1) framerate is king
        2) getting rid of advanced lighting, bump mapped animated textures, smoke, fog, clouds, falling snow, rain, etc, etc make your opponent easier to spot.

    • No discussion of how many blades there are on razors these days is complete until somebody posts this [theonion.com]. So I figured I'd just get it out of the way :-)
  • by Anonymous Coward on Thursday April 10 2008, @01:59PM (#23028136)
    How many power supplies are required? Does it come with a 12 KV step-down transformer and 220V three-phase power hookup? Can I heat the basement with it?
  • The end of the article makes a good point. While we have dual- and tri-core graphics setups now, the programs are not designed to exploit them. This is the same issue that is being faced in the general CPU market as well. If you don't have a multi-threaded app for multiple CPUs, you only gain in multitasking, but not in a single program. A serial program can ONLY run serially. There's only so much parallelism that a CPU can infer. And at that point, you have out-of-order execution to make up for mutli
    • Actually not a bad idea, quite a bit of objects are stationary in a game, so you can offload that to secondary GPU's - the problem is most games/engines haven't been designed for this in mind and you can't magically create something that can do this for them.

      Over time the GPU's will be more flexible and that means it will be easier to offload calculations through some common API, but I think it will be a few years yet before this potential can be realized.
  • by default luser (529332) on Thursday April 10 2008, @02:04PM (#23028198) Journal
    It's not necessairly a limit of the board design, but a limit to what game engines can be optimized for. Most game engines do not scale well beyond two cards, as can be seen here:

    http://www.xbitlabs.com/articles/video/display/zotac-9800gx2.html [xbitlabs.com]

    While there are a few key games that get no boost out of 2-way SLI, the vast majority of games do see improvement. 3-way, on the other hand, can actually cause WORSE performance.

    It probably has to do with limitations on how the SLI/Crossfire drivers can fake-out the game engine. There are probably limits to how many frames the game engine allows to be in-flight at once, limiting how much performance boost you can get from AFR SLI. And although you can get around game engine limitations with split-screen rendering, this mode needs specific game support, and shows less potential performance increase. Plus, split-screen rendering and has to be selected explicitly in Crossfire (AFR is the default).
    • Re: (Score:2, Funny)

      by Anonymous Coward

      3-way, on the other hand, can actually cause WORSE performance.
      Please don't tell my girlfriend that, I've been working on her to do that with me for months now.
    • Split-screen SLI also causes problems with deferred rendering and other post-processing techniques that have become all the rage these days - the driver has to work around the fact that none of the chips has the complete framebuffer.
  • by Yvan256 (722131) on Thursday April 10 2008, @02:31PM (#23028502) Homepage Journal
    Why can't we put our efforts toward more efficient GPUs? Just as most users won't ever be able to push their current CPUs at their maximum, most aren't even using the full power of their GPUs.

    I want a fanless, 5W GPU with the power of GPUs from about 3 years ago. Can the new smaller transistors allow for this or am I asking for too much?

    If ATI and nVidia keep pushing for raw power, they'll get beaten to the low-power finish line by the likes of intel and VIA.
    • Re: (Score:3, Informative)

      You're not asking too much, you're simply over-valuing what 3-year-old tech is capable of. The GeForce 6800 Ultra was the best Nvidia card in existence 3 years ago, and soon, you will be able to purchase the 9500 GT, which will have more performance.

      The card features %20-30 more performance than the 8600GT (plenty to top GPUs from 3 years ago), and with a 65nm process, should consume around 30w or less at-load.
    • I want a fanless, 5W GPU with the power of GPUs from about 3 years ago. Can the new smaller transistors allow for this or am I asking for too much?

      Top of the line? Then no, they were eating ~|00W then and would still be eating 25W even if we inverse-applied Moore's "law". However, non-gamers look to be in for a treat, for example the Atom's chipset does HD decoding at 120mW. Yes, it got some cripplings but say within 0.2W it should do full HD. Gamers are going to be pretty alone with their power rigs in not that many years...

  • by sanosuke001 (640243) on Thursday April 10 2008, @02:33PM (#23028540)
    The FA states that multi-threaded gaming is fundamentally flawed. How is this a valid statement? They are testing a multi-cored GPU using games that were most likely only developed to use, at MOST, two cores. Regardless of how many core you throw at it, the application (ie. your game) will NEVER use any of those cores.

    In fact, gaming and graphics scale amazingly well as a multi-threaded application. In fact, as many in the graphics/gaming community have been stating recently, ray tracing would benefit greatly from more GPUs. Being able to trace multiple rays at a time would speed up rendering.

    They state that it is fundamentally flawed when they should have said that it would be ignorant to assume that an application designed to use a single-core or dual-core GPU would benefit from extra GPUs.
  • Parallelism (Score:3, Insightful)

    by kvezach (1199717) on Thursday April 10 2008, @02:35PM (#23028562)
    As opposed to raytracing, which is so extremely parallel [wikipedia.org] that it scales nearly linearly. In other words, your 3-way real-time raytracer graphics card (if/when such a beast is ever made) would perform at about 2.8x the one-GPU variant. And unless the Rapture^Wsingularity keeps you from getting a 192-GPU card, it'd render at 170x the reference or so.

    (Of course, there's the question of global illumination. I don't know if those can be parallelized as easily, but there was a story about distributed photon mapping here some time back, where they used Blue Gene.)
  • I wonder if their stategy discussions follow this tone:

    http://www.theonion.com/content/node/33930 [theonion.com]
  • Imagine a beowulf cluster of these!
    • Wow! A beowulf cluster of graphics cards! That certainly ought to be enough to run Duke Nukem Forever!
  • FYI (Score:3, Insightful)

    by geekoid (135745) <dadinportland.yahoo@com> on Thursday April 10 2008, @03:26PM (#23029248) Homepage Journal
    " so it's not going to see release at any point in the future,"

    The future is a really long time.
  • by British (51765) <british1500@gmail.com> on Thursday April 10 2008, @03:52PM (#23029562) Homepage Journal
    "Fuck everything, We're doing three GPUs!"
  • by SparkleMotion88 (1013083) on Thursday April 10 2008, @05:42PM (#23030564)
    This is a terrific advance in the field of cramming. I'm looking forward to seeing their presentation at the Cramming and Stuffing conference later this year.
  • by seeker_1us (1203072) on Thursday April 10 2008, @08:16PM (#23031724)
    http://en.wikipedia.org/wiki/Voodoo2 [wikipedia.org]

    The Voodoo2 was a set of three graphics processing units (GPU) on a single board, made by 3dfx. It was released in February 1998 as a replacement for the original Voodoo Graphics chipset.

    :)