ATI stakes claims on physics, GPGPU ground

One of the more surprising aspects of ATI’s Radeon X1000 series launch is something we didn’t get a chance to talk about in our initial review of the graphics cards: ATI’s eagerness to talk about using its GPUs for non-graphics applications.

ATI practically kicked off its press event for the Radeon X1000 series with a physics demo running on a Radeon graphics card. Rich Heye, VP and GM of ATI’s Desktop Business Unit, showed off a simulation of rolling ocean waves comparing physics performance on a CPU versus a GPU. The CPU-based version of the demo was slow and choppy, while the Radeon churned through the simulation well enough to make the waves flow, er, fluidly. The GPU, he proclaimed, is very good for physics work, and he threw out some impressive FLOPS numbers to accentuate the point. A Pentium 4 at 3GHz, he said, peaks out at 12 GFLOPS and has 5.96GB/s of memory bandwidth. By contrast, a Radeon X1800 XT can reach 83 GFLOPS and has 42GB/s of memory bandwidth.

This demonstration was more important as a statement of position and direction for ATI than anything else. NVIDIA has been rather enigmatic about any plans for physics processing, and has seemed to step out of AGEIA’s way for the most part, welcoming the PhysX effort as another means of improving PC gaming. ATI is clearly ready and willing to take on AGEIA’s PhysX card by using its GPUs for physics simulation, and the company believes that the more general programmability of its Shader Model 3.0 implementation in the Radeon X1000 series could make it a viable competitor there. There was talk of a CrossFire dual-GPU config in a split setup, with one GPU handling graphics and the other handling physics for certain games, and somebody present even suggested the possibility of a third GPU on a PCI card dedicated to physics processing. The folks from ATI seemed to like this suggestion.

We haven’t heard many specifics yet about how ATI plans to go about exposing the physics acceleration capabilities of its GPUs to developers. One obvious path forward would seem to be cooperation with physics middleware vendors like Havok, but AGEIA has already made extensive inroads into next-gen games development by giving away licenses for its PhysX API. If ATI is serious about making this push, it may have to slog through an API war by pushing developers to use something other than PhysX API. We shall see.

Beyond just physics, ATI is happy to see its GPUs being used for all sorts of general-purpose computation. To that end, it invited Mike Houston from Stanford to come talk about his efforts at GPGPU, or general-purpose computation on graphics processing units. Houston gave a nice overview of how graphics lingo translates into general purpose computation, where textures are used as general storage and pixel shaders operate on data other than pixels. There’s a PDF of his talk available online, so you can read it for yourself. Houston particularly liked the Radeon X1800’s new features, including 32-bit floating-point precision, long shader programs with flow control, and faster data uploads and downloads over PCI Express. He also praised the X1800’s large number of available registers and its support for 512MB memory, which he said is sorely needed in GPGPU applications.

Houston gave several examples of applications where GPUs can outshine CPUs for certain types of data-parallel processing. One of the most exciting was an implementation of the GROMACS library that’s used for protein folding, as in Folding@Home. Although the GROMACS implementation wasn’t yet optimized for the Radeon X1000 series and still used Pixel Shader 2.0b, the Radeon X1800 XT was already performing between 2.5 and 3.5 times faster than a Pentium 4 3GHz processor. The GeForce 7800 GTX didn’t fare so well, achieving only about half the speed of the P4 3GHz. Houston offered a laundry list of things that GPGPU developers need from graphics chip makers, including better information about how to program the GPU and more direct access to the hardware.

To that end, ATI pledged to open the X1000 platform to developers by documenting the GPU architecture for use in data-parallel processing and providing a thin layer of abstraction so GPGPU developers can get more direct access to the hardware. ATI also said it planned to “seed developers with tools and prototype applications” for X1000 development.

Comments closed
    • Amun
    • 14 years ago

    Here you go guys, a 1800xt for $109 bucks!!!

    §[<http://www.pricegrabber.com/search_getprod.php/masterid=11173272/<]§

    • dlang
    • 14 years ago

    with all this talk about an open API to use the card for physics processing I hope they also open the API for regular graphics use as well (i.e. for makeing Linux drivers)

    • Tupuli
    • 14 years ago

    The whole GPGPU thing is a waste. Even fairly trivial things like conjugate gradient (a common step in solving PDEs like navier-stokes, or elasticity) aren’t terribly fast on the GPU despite it’s amazing peak FLOP performance.

    By the time GPUs have the generality to do these sorts of numerical algorithms we’ll have 8 core CPUs.

    Notice that they are comparing top of the line cards with careful optimization (i.e. painstaking implementation) versus a 3Ghz Pentium. Why not compare against a dual core chip that can accelerate a broad class of problems? Even the Xbox will have 3 general purpose processors.

    The only reason that the GPU has such amazing performance is its dedication to a specific task. It will never have a reason to have good branch or cache performance, things that are very important for good performance in numerics.

    Unless the GPU can outpace the CPU by a factor of 10 or more, no one will bother implementing these algorithms on the GPU. Far more likely is a low-power, general, multi-core processor.

    I see lots of hype (the GPU is a supercomputer!), but little substance.

    • quarantined
    • 14 years ago

    “Although the GROMACS implementation wasn’t yet optimized for the Radeon X1000 series and still used Pixel Shader 2.0b, the Radeon X1800 XT was already performing between 2.5 and 3.5 times faster than a Pentium 4 3GHz processor.”

    Wow. Paper launches aside, all of this halfway redeems the X1000 generation in my book. Hopefully this ground work will lead to an impressive next generation for ATI.

    • shank15217
    • 14 years ago

    Oh there goes another use for CPUs hehe. We’ll see in about 2 years dedicated Crossfire GPU systems that work with distributed computing APIs. I got $5 on that!

    • Grigory
    • 14 years ago

    Oops I meant to reply.

    • Skirrow
    • 14 years ago

    The cool thing about the ATI gpu being able to do physics processing is this.

    Picture the scenario.

    Its 2007. Your X1800XT is getting a bit long in the tooth now and has trouble with the latest games but ATI or Nvidia have just released their new cards with all the new gizmos and you decide to upgrade. So you buy a nice new card and enjoy smooth gameplay once again.

    But wait. What to do with the trusty old X1800XT?. Then you remember ATI’s drivers have support for their physics engine, GravATI (see what i did there? i bet they end up using that name :D) which allows cards to do physics processing.

    Saves having to fork out for a new physics card AND your old card wont need throwing out.

    Good news in my book. I’ve got a drawer full of old cards i’m holding onto only because i sunk a hell of a lot of cash into them.

    • bthylafh
    • 14 years ago

    $5 says there’ll be a physics API in DirectX 10 or 11, and we’ll have a situation similar to video cards — some are more capable than others, and the stuff that h/w can’t handle will be done on the CPU.

    I hope there will be an open cross-platform API for physics, much like OpenGL.

      • Ryszard
      • 14 years ago

      Microsoft are actively hiring for DirectPhysics.

    • ArturNOW
    • 14 years ago

    But you need 2 graphic card to do physics. One render scenes and the other one handles physics, right? So all in all its cheaper to buy PhysX and this card needs less power…

      • Flying Fox
      • 14 years ago

      Well, those 512 threads in the R520 aren’t just for show. Besides, I would think they can dynamically allocate a quad (or 2) for the physics stuff? Of course, using a 2nd card would be the best but I don’t see why it cannot be done with 1.

        • Usacomp2k3
        • 14 years ago

        Seems like it would use up resources (especially memory) that the textures could make use of. Thus, it’s not, in a sense, “free”

          • mrzeld
          • 14 years ago

          in current cards it would be a “waste” of resources. but imagine when our desktop cards are more like the XBox 360 GPU. 48 (or more) general purpose piplines.

            • ArturNOW
            • 14 years ago

            That’s what i’m talking about 🙂

    • indeego
    • 14 years ago

    This article needs reference to the creamy smoothness of rolling waves of virtual butterg{<.<}g

      • Grigory
      • 14 years ago

      snack time

    • Shinare
    • 14 years ago

    So this is how ATi is going to sell their somewhat lackluster new line of GPUs….

      • wierdo
      • 14 years ago

      Lackluster is stretching it, call it the second best you can buy for maybe two thirds of new games.

      This physics processing thing is nice, but it wouldn’t be a factor in my buying decision for another two years at least I would guess.

      Update: Nice review at §[<http://www.behardware.com/articles/592-1/ati-radeon-x1800-xt-xl.html<]§ It seems there's more to this card than I first thought. Good potential.

    • sigher
    • 14 years ago

    Well shader model 4 requires the pipes to be multipurpose and long instruction lines etc, so both the next nvidia as well as ati will have a lot of possibilities, the only thing with nvidia though is that they are not very forthcoming to developers about such details of the hardware, you see that with the ‘video accelerator’ thing on the nvidia cards, I only see crappy WMP-DRM enforced acceleration and no independant developers adding it, due to lacking information I assume.
    So hardware-wise it’s possible on nvida and very possible on future nvidia cards I think, but due to lacking support I don’t expect anything but a vague possibility of nvidia hammering out a deal with aegis and using their api through the closed source drivers of nvidia.
    Whereas ATI opened the door now it seems, I applaude them for their care.

      • Krogoth
      • 14 years ago

      SM 4.0 hasn’t been iron out yet. “DX10” will be released as soon as Vista gets here.

    • drfish
    • 14 years ago

    Welcome to Sunday afternoon Scott… 😉

    §[<https://techreport.com/forums/viewtopic.php?t=34399<]§ All that other stuff is great and all, but I really only care about the folding... Bring on the GPU client!!! :D ~25Ghz of folding power in one system... *drool*

      • Krogoth
      • 14 years ago

      The big problem with a GPU client is the difficulty of optimizing code on various range of architechs with new ones being introduce at a much faster pace then CPUs.

    • Thrashdog
    • 14 years ago

    So now I wonder if NVidia is going to get after AGEIA to put PhysX chips on their graphics board to compete?

    • Sargent Duck
    • 14 years ago

    Heh Heh Heh……

    I was thinking to myself, ’bout 2 weeks ago, something like this:
    “Maby the reason the r520 is so delayed is because ATI has put a physx processor on there without telling anyone”. Heh, kinda wish I had posted that, maby even put some money on that……

    If indeed the r520 is this multitalented, it may explain the higher price. Of course, this all matters little when comparing framerates with the 7800gtx.

      • DrDillyBar
      • 14 years ago

      Playable frame rates AND ~1000 pts / day.

      • Shining Arcanine
      • 14 years ago

      They are most likely using DirectX and OpenGL instructions more than anything else, meaning the Radeon X1000 series is no more or less programmable than Nvidia’s GeForce 6 and 7 series. They might have made a few hardware/driver optimizations for things that games don’t use but this type of processing would use through.

    • Tairc
    • 14 years ago

    I see a lot of potential here, especially if the interface/API’s get hammered out ASAP. My biggest hope would be that Havok/AGEIA get in gear and work out an API licensing agreement with ATI, allowing the same general API to do physics on both types of hardware. Of course, the ATI card isn’t pure hardware like the AGEIA card, and would be doing graphics, but it would allow acceleration (get this) even if you didn’t buy an AGEIA card. This provides a very nice upgrade ‘path’ for the average user that doesn’t want to spend 200$ on a separate physics card. That is, until he sees games supporting physics, because all the ATI cards have physics support on them as well.

    To me, giving ATI the means to support your API is a great way to get developers to recognize that there’s going to be a large enough user-base to warrant development. Sure, you’ll lose a few sales because somebody’s X1000 is going to do their physics for them – but in a year, when more and more games require both bigger graphics cards AND more physics, suddenly, you can choose to either buy a new 500$ video card, or simply buy a 200$ physics card. The physics card will offload the physics from your GPU, improving both playability and framerate.

    I’m just saying, is all.

      • robg1701
      • 14 years ago

      That idea has merits, but it also means you start getting into an area of different additional physics processing capabilities for each computer, which could complicate matters for developers when using the addons. So to some extent, if AGEIA were to maintain a stranglehold on PPUs it could be better for the devloper having a constant unit to target.

Pin It on Pinterest

Share This