Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
Hardware

World's First Physics Processing Unit 494

Duane writes "Gamers Depot has an exclusive interview with the team behind Ageia - the maker of the world's first Physics Processing Unit (PPU) - which was just announced today. "Sure we've all heard about the CPU and GPU - that's old hat by now and as most hardware reviewers will tell you, it's about time we got something that's truly revolutionary. Yeah, Pixel shaders are cool, and can do a lot of really nice things; however, pale in comparison in scope to what the PhysX chip from Ageia has the potential to bring to gaming.""
This discussion has been archived. No new comments can be posted.

World's First Physics Processing Unit

Comments Filter:
  • by stupidfoo ( 836212 ) on Tuesday March 08, 2005 @03:48PM (#11880259)
    All I know is that I want to throw the dead hooker down the stairs and have her head split open... or whatever that anti-violent game ad says I can do.
  • by Nom du Keyboard ( 633989 ) on Tuesday March 08, 2005 @03:48PM (#11880261)
    It won't be /. worthy news until the first Linux port is up and running on it!
  • by imstanny ( 722685 ) on Tuesday March 08, 2005 @03:48PM (#11880264)
    Nerds around the world rejoice!
    • by GillBates0 ( 664202 ) on Tuesday March 08, 2005 @03:50PM (#11880290) Homepage Journal
      PPU: Pr0n Processing Unit.
    • by bonch ( 38532 ) on Tuesday March 08, 2005 @04:47PM (#11880988)
      We've already pushed off sound, graphics, and now physics onto seperate processors. Why not just craft an entire game console onto a single card and be done with it? Jeesh.
      • by Surt ( 22457 ) on Tuesday March 08, 2005 @05:24PM (#11881404) Homepage Journal
        A game console doesn't have one of these (yet), nor are even the next generation likely to.

        What this is suggesting is rather that games are for the most part not general purpose tasks, and that as a result general purpose cpus can be grossly outperformed by special purpose cpus. Once you reach that notion, then you just have to decide what the set of special purpose cpus you need are. It's a repeating process where parallelizable areas of the codebase are identified, and special purpose cpus are crafted to handle them, so that the performance limiting area of code keeps moving to some task for which the special purpose chip hasn't yet been built.

        For quite some time the graphics capabilities of the GPUs has been the limiting factor in effectively conveying the game designer's intended experience. We're now reaching the point where the GPUs are so effective that what now looks 'wrong' has more to do with physics simulation than with graphic rendering. (Though I'll still say that there are 3 or 4 generations of graphics improvements yet to come that will still have a significant effect, it's just that now it has reached the point where it is no longer clear that more GPU improvements will have the _largest_ effect on perceived quality.)

        • "A game console doesn't have one of these (yet), nor are even the next generation likely to."

          I don't get what is so different between a GPU and this PPU thing. A GPU is mainly multiplying vectors and matrices and dot products and division, physics simulation is not very different, although the result is handled differently. As far as current consoles, the PS2 VUs are able to handle a lot of physics related tasks, and with the CELL in PS3 I'm hoping for even more capable physics handling. I'd like to see mo
          • I don't get what is so different between a GPU and this PPU thing. A GPU is mainly multiplying vectors and matrices and dot products and division, physics simulation is not very different, although the result is handled differently.

            Quite differently. You are suggesting that one draws the matrix results and the other just stores the matrix results, but there is a more important factor than that. All the data that is pushed onto the grpahics card is essentially on a one-way trip. After going through the T
  • by Anonymous Coward on Tuesday March 08, 2005 @03:48PM (#11880267)
    Go outside.
  • by eno2001 ( 527078 ) on Tuesday March 08, 2005 @03:49PM (#11880278) Homepage Journal
    ...that my suspcions were correct. All this 3D stuff with pixels and texels and blah blah blah is just test runs before we create physical augmentation with nanotech replacing pixels and texels. (Wow that was one sentence!) Holodeck anyone? ;P
    • by Perl-Pusher ( 555592 ) on Tuesday March 08, 2005 @04:21PM (#11880678)
      Upgrades to continue as long as we continue to gain knowledge and make new discoveries. Can't guarantee a ship date though. With the Hiesenberg UP module the enemy can't do anything until sighted, detected or measured. Stand still, close your eyes, shoot, winner everytime.
    • Re:More Proof... (Score:5, Insightful)

      by nine-times ( 778537 ) <nine.times@gmail.com> on Tuesday March 08, 2005 @04:22PM (#11880704) Homepage
      More seriously, it does seem that the video game industry has been moving more and more towards complete world simulations rather than "games with rules". Maybe that's an obvious statement, but part of the reason I say that is that it's not a necessary motion in the industry, but a consumer-driven one. In other words, developers could keep trying to innovate on the Super-Mario-type games, but gamers and developers seem more focused on creating more realistic first-person-shooter war simulations. (Not that I'm criticizing)

      Anyway, what I'm getting at is that a holodeck-like experience does seem to be what both gamers and developers have set up in their minds as the "holy grail" of video games. I think in the near future, we're going to see real innovation in physics engines to use ray-tracing-like lighting affects and real particle collisions instead of the pre-programmed tricks used today. I think for the transition we're in for, it probably would be appropriate to compare the transition to the sort of change we saw between the fake 3D of Duke Nukem 3D to the [more] real 3D of Quake.

      However, what remains to be seen is whether those games will be more fun.

      • Re:More Proof... (Score:3, Insightful)

        by edwdig ( 47888 )
        In other words, developers could keep trying to innovate on the Super-Mario-type games, but gamers and developers seem more focused on creating more realistic first-person-shooter war simulations. (Not that I'm criticizing)

        Try playing console games and you'll see things shifted much more the other way. It just comes down to First Person Shooter games play much better with a keyboard and mouse than with a controller, whereas a classic Nintendo style game needs a fairly well defined controller to play well,
  • ...right here [ageia.com]. It doesn't really say anything, though - just a few pages that recap physics usage in games, and then a paragraph about how they're going to change all that, etc.

    Didn't white papers use to be heavy on technical content? Now it seems that "white paper" just means "nicely formatted eight page PDF advertisement"....
  • by Anonymous Coward on Tuesday March 08, 2005 @03:50PM (#11880288)
    Note that no-where in the press release does it say that this is a shipping product. Before you get all excited about the promise of this product, realize that this chip may never see the light of day. A press release does not a product make, regardless of how cool the product might be.
  • Wow! (Score:4, Funny)

    by cavemanf16 ( 303184 ) on Tuesday March 08, 2005 @03:50PM (#11880295) Homepage Journal
    And I thought Slashdot "editors" had poor grammar skills! Damn. I guess they're starting to farm the technical report writing and gaming reviews to India now too!
  • Interesting idea (Score:5, Interesting)

    by Dark Paladin ( 116525 ) * <jhummel@nosPAM.johnhummel.net> on Tuesday March 08, 2005 @03:52PM (#11880327) Homepage
    I just finished reading the article, and this actually has some potential.

    The biggest problem they're going to have to deal with, and granted, I'm not a game developer so someone can feel free to fill in the details, is that I would believe that most developers have their own method for dealing with physics - from simple collision to ragdoll and the like. The idea is "How do I tell the computer these things are touching each other' (like bullets - these are "instant shot", so the developer just says "if there's a straight line between the direction the Player A is facing, and if that line would intersect Player B, then it's a hit. If not, then miss." And algorithms like that are done by matrixes, if I'm not mistaken. Other "hits" deal with actual objects (rockets moving, goops from the goop gun, etc).

    But the difference between Quake III and Unreal Tournament is more than just 'draw the graphics", it's also in how each engine deals with how those collisions are managed.

    So with a PPU, you have to decide on a common library of collisions. Good news: more objects you can play with and let the PPU decide what's getting hit. Bad news: everybody's game will react basically the same and they'll have to decide if that's a good idea.

    Either way, I'll wait a year or so and see what happens. Best of luck to the developers - looks like they're at least shooting for something unique.
    • by Anonymous Coward on Tuesday March 08, 2005 @04:02PM (#11880460)
      I just finished reading the article, and ...
      You must be new here.
    • So with a PPU, you have to decide on a common library of collisions. Good news: more objects you can play with and let the PPU decide what's getting hit. Bad news: everybody's game will react basically the same and they'll have to decide if that's a good idea.

      It looks like we're heading there already. Havok [havok.com] has already developed a mature software physics engine which is used in many popular games. I think in this case, developers are willing to give up a little control on physics to have better looking ef

    • by RichardX ( 457979 ) on Tuesday March 08, 2005 @04:06PM (#11880509) Homepage
      Not really.
      You could have said the exact same thing about graphics with the advent of hardware 3D accelerators, yet games certainly haven't all ended up looking the same. If anything they're able to look *more* varied now thanks to the extra power allowing neat tricks like cell shading and real time effects.

      In the same way GPUs (initially, at least) sped up all the graphics things that all 3D games have in common (triangles, texturing, lighting, etc), this will presumably speed up all the physics things all games have in common (collisions, velocities, etc)

      That doesn't necessarily mean they all have to act the same. As a programmer you still get to determine exactly what happens when something collides, or how it behaves when it's crushed. It's just that you have access to much more power, and in the same way that gets us neat tricks on GPUs I think we'd see the same with these PPUs.

      The important thing is that this takes care of all the low level stuff, giving the developers more time and power to spend on the higher level areas where they can really be creative.

      Incidentally, am I the only one here saying "about time" with this? I had this idea the moment I saw the first Voodoo card. I'd have done something with it, but I figured it was so damn obvious everyone else would've thought of it too. That, and I'm just plain lazy :)
      • Incidentally, am I the only one here saying "about time" with this? I had this idea the moment I saw the first Voodoo card. I'd have done something with it, but I figured it was so damn obvious everyone else would've thought of it too. That, and I'm just plain lazy :)

        Yea I can relate to that very well. When I started playing Quake II and Unreal the other thing I wanted was a small keyboard, and a mouse/keyboard just for playing those FPS games faster. Now they have them.
    • Re:Interesting idea (Score:4, Informative)

      by mikael ( 484 ) on Tuesday March 08, 2005 @04:08PM (#11880534)
      I would believe that most developers have their own method for dealing with physics - from simple collision to ragdoll and the like.

      Basic collision detection methods are bounding planes, spheres, capsules, and axis-aligned boxes, along with Binary Space Partitions, Quadtrees and Octrees combined with particle systems. It would be fairly straight forward design an instruction set to perform these operations between the simple primitives (spheres, planes). But BSP Trees, Quadtrees and Octrees would require a high level data format.

      If all the collision testing could be done within a single thread within the time limit of a single frame, it would be no different from the player-missile and sprite graphics implemented on early home PC's (Atari computers could do hardware base per-pixel collision detection). Although, it would probably seem easier to have additional vector processors like Sony's Cell processor.
    • As a loyal slashdot reader, I didn't go anywhere near the article, nevertheless, I feel that for a PPU to be successful in the marketplace, it would need to have a certain level of flexibility. Not only would things like local gravity, scale, real-time vs slo-mo need to be taken into account and be modifiable, but also, you'd need to be able to allow the developer to bend/break the rules at will to account for magic, powerups, etc.

      BTM
    • You're kind of hinting on something I've been wondering: what is the API for this? There's nothing currently out there like OpenPL (Open Physics Language); they apparently have this proprietary thing called Rocket and NovodeX, but unless there's a standard way to talk to cards (and have others make cards to really get innovation going), this will take some time to gain traction.

      I also have to say I'm a bit miffed because I've had a similar idea floating around in my brain for a while. I really need to star

      • Re:Interesting idea (Score:5, Informative)

        by Atzanteol ( 99067 ) on Tuesday March 08, 2005 @04:27PM (#11880765) Homepage
        Mmmmm. Apparently there is an open source lib for physics.

        http://ode.org/ [ode.org]

        But your point about a standard like OpenGL not existing is true. We'll probably have a rehash of the early graphics library incompatabilities again. ::shudder::

        You think people would learn. Open standards help a new technology to expand and to become accepted. It helps *everybody* in the industry.
    • Physics processing boils down to one thing:

      INTERSECTIONS (sometimes known as collisions)

      If all solid objects are rendered as sets of triangles, it is conceptually simple to have the physics engine report back which triangles are intersecting each other. Ideally the engine is sophisticated enough to report when a triangle from set A intersects with a triangle from set B (where A and B might be the set of triangles that make up a player and the set of triangles that make up a rocket, respectively).

      Determin
      • Re:Interesting idea (Score:3, Informative)

        by Have Blue ( 616 )
        Not really... Physics processing boils down to *2* things, which are themselves huge and complex fields:
        • Collision detection. You have some representations in the world- are any two of them intersecting? The answer to this depends on how they are represented and can be fast (boxes, cylinders, spheres) or expensive (polygon soup) or anything in between. The triangle algorithm you describe is indeed simple, but it's usually not feasible to run it on every triangle pair in the world in every frame- which is w
    • Re:Interesting idea (Score:2, Interesting)

      by genneth ( 649285 )
      Actually, it wouldn't require the physics to be the same. Just like graphics are not the same between Doom 3 and an X server using GL for accelerated window drawing, they can both be sped up by OpenGL. For example, one of the biggest "sinks" for CPU power with physics simulation is collision detection. The algorithm for collision detection between (N-dimensional) polyhedral objects, optimised with interframe computation re-use, is already very well known -- it's the Lin-Canny algorithm, and manages amortize
    • What if you're playing multiplayer with different PPU client hardware? That would seriously mess up gameplay if collision/arc/timing/etc calculations are off.
    • the way i understand it, the chip will have physics as real as possible, and since it's in hardware it's (hopefully) blazingly fast.

      the reason engines deal differently with physics right now is a matter of optimisation. different developpers have different ideas on how to optimize (or sometimes they don't...) with different resulting speeds.
    • by joib ( 70841 ) on Tuesday March 08, 2005 @04:35PM (#11880860)
      Normal cpu's already have pretty good fpu units, which are very fast for scalar code. Also, we have things like SSE2/Altivec for vectorizable code. And then there's things like gpgpu.org looking at using the massively parallel fpu capacity of modern gpu's for general purpose physics calculations (linear algebra), i.e. vector processing on a budget.

      So where does this thing fit in? As expected, the "article" was nothing more than a thinly veiled marketing blurb, so no info there. Personally, I find it hard to believe that the PPU is competetive with FPU's and GPGPU for general purpose FP calculations. That leaves a chip optimized for certain operations, a bit like MDGRAPE. Or what am I missing?
    • Re:Interesting idea (Score:5, Informative)

      by BannedfrompostingAC ( 799263 ) on Tuesday March 08, 2005 @04:43PM (#11880947)
      like bullets - these are "instant shot"
      Wrong. Have you ever fired a gun? It actually fires a gyroscopically-stabilized projectile that takes a discernable amount of time to reach its destination. Hitting a non-stationary object reliably at long range (800m-1000m) is next to impossible.

      This matters at the physics level. If you are going to fully implement the ballistics you are going to have implement the motion of the bullet, the atmospheric drag on the bullet, the gyroscopic stabilization, the effect of gravity on the bullet ("bullet drop") not to mention the effects of the individual specifications of the bullet itself, and perhaps some entirely random factors (the world isn't perfect).

      And if you are implementing a game where players can fire an assault rifle full-automatic (600-700 rounds a minute or more, depending on too many factors to list - which might need to be implemented and calcuated by the computer, of course...) you can see that the CPU is going to start needing some help to work it out.

      And that's just the bullets.

      The gun example is just an example of the sort of jobs a co-processor might be required to do in an FPS environment. To cut a long story short, if you are going to be simulating life, even a small approximation of life, accurately, you are going to need to be calculating an awful lot of physics.
  • by TheNecromancer ( 179644 ) on Tuesday March 08, 2005 @03:52PM (#11880328)
    This includes things such as Rigid Body Dynamics, Collision Detection, Fluid Simulation, Soft Bodies and Fracturing of objects.

    This will be useful for all those pr0n sites out there!
  • by ivan256 ( 17499 ) * on Tuesday March 08, 2005 @03:53PM (#11880341)
    ...but coincidentally the 1,000,000,000th computer accessory that will be a complete failure in the market.
  • by markmcb ( 855750 ) on Tuesday March 08, 2005 @03:53PM (#11880352) Homepage
    If this thing can do physics homework, I'm getting two.
  • So which physics model are they using? I'm pretty sure it's not the RealLife(tm) one. That being the case how the hell are they going to upgrade it.

    We have some very sophisticated software for newtonian physics modelling. We already have very fast general purpose hardware. Is this add on any more than just neon lighting for gamers?
  • Yet another reason for Duke Nukem Forever to be delayed.
  • by wcrowe ( 94389 ) on Tuesday March 08, 2005 @03:57PM (#11880400)
    Considering that most games routinely defy the laws of physics, I would think that such a processor would actually make the games more dull.

  • by Deep Fried Geekboy ( 807607 ) on Tuesday March 08, 2005 @04:00PM (#11880432)
    If you have a game like Unreal Tournament 2004, it is the physics processing that really kills your framerate, no matter how good your GPU. You can see this by simply swapping between the Deathmatch and Onslaught gametypes. The Onslaught world is filled with vehicles which run off the Karma physics engine, and they KILL your framerate, so that the game effectively becomes CPU-throttled, instead of GPU-throttled (which is what we are used to). A PPU is a genuinely brilliant idea, and relatively easy to implement. It will be interesting to see what the programming interface is... and whether the board runs an engine like Karma or something they've invented all for themselves. Prepare to be amazed, I think.
    • What makes you think this isn't killing the graphics side of things?

      If you look at most games they have static worlds, so why don't they have move moving parts like doors, windows, fans, windmills, fixed physics holes blown in walls even if there's no physics attached to them?

      Because move able objects kill all you attempts to optimise the 3D objects, BSP doesn't work for deformables, QTrees can be done, but it's quite intensive and you have to use non-uniform QTrees which are a lot slower.

      So for a PPU to
    • by clutch110 ( 528473 ) on Tuesday March 08, 2005 @04:42PM (#11880936)
      Here [gamespot.com] is a link to an Epic developer talking about how the Unreal 3 engine will use this PPU.

      Very interesting technology, comes with its own SDK and should be able to handle many times the amount of physics based objects in a game than the CPU can handle now.


  • "What is increasingly defining successful games is how well they emulate reality," said Rob Enderle

    Now seriously, why would I want to play a game where I need to sit down and rest after running 3 flights of stairs?
  • I Wonder... (Score:3, Insightful)

    by Nom du Keyboard ( 633989 ) on Tuesday March 08, 2005 @04:02PM (#11880462)
    I wonder which unannounced, next generation game console would jump on this.

    I also wonder how it compares to the Cell processor's dedicated units.

    For a console, sounds like someone could really steal a march on the rest of them...

    ...Or become the next Amiga.

  • I like the idea of bringing Physics calculations to a seperate processor (then again, I'm a noob on hardware) ; as physics seem to be of great importance to bring quality interactivity to a user/gamer.

    Whereas Havok (http://www.havok.com/) seems to be doing a great job , software wise, I don't find it too strange to let those calculations be done on a seperate unit ; and take the strains/limits off that we have with current hardware.

  • Hopefully once these fall into common usage they'll re-release the "classics" such as Dead or Alive Beach Volleyball. You know, games that could really benefit from this technology.
  • by _ph1ux_ ( 216706 ) on Tuesday March 08, 2005 @04:05PM (#11880499)
    I think it would be interesting to PPUs in use in robotics. Maybe these devices can give better balance sensors - or provide intrinsic abilities for robots to know/sense how to navigate and interact within the world.

    Shouldnt mechs use this to create highly mobile bipedal motion with a good ability to balance in chaotic environments (fast paced combat)

    ???
  • by turgid ( 580780 ) on Tuesday March 08, 2005 @04:05PM (#11880503) Journal
    but what the heck is it? Is it just a RISC processor with lots of FP SIMD units for doing lots of sums in a hurry? Is it VLIW? Is it related to any existing CPUs? Is it just the next evolutionary development of the current generation of GPUs?

    Pictures of boards are all well and good, and the martketing hype is fun, but we need to know.

  • by TomorrowPlusX ( 571956 ) on Tuesday March 08, 2005 @04:06PM (#11880516)
    I'm not a professional developer... that said I'm developing both a robotic simulation API/framework and a game, both in my free time, both *heavily* use the open dynamics engine for physics.

    The Open Dynamics Engine is free, & open source. It's not the best physics engine, by any margin. However, being open source I can afford it... and most importantly I can use it on my Mac ( hell, I actually provided some patches to get it to correctly use single-precision trig when OS X.3 came out ). Plus, I want to release my game and robot simulator under an open source license... can't expect people to *buy* novodex or havok just to build the apps.

    This PPU looks like a *wonderful* thing, but reading their site, and the interview, it sounds like to use it you've got to use Novodex. That said, Novodex is awesome -- and many games use Novodex already for physics.

    (Perhaps I missed something, maybe Novodex is just an API wrapper. Maybe they'll have a low-level API which you can bind to as you want. )

    But the thing is, I'd like to be able to buy one of these boards and *not* have to shell out for a developer license for an API which isn't even available on Mac ( maybe it is ). Also, both my simulator and game are intended to be released under an open source license at some point. So, no novodex for me. So, no PPU for me.

    Perhaps we're just a little short on data at the moment.
    • I think, to clarify my point above, it should be pointed out that you and I can right now write open source games using OpenGL or DirectX; anybody can get the SDKs ( I don't know what dev tools cost on windows, but for Mac OS X and linux it's free ).

      So, my question is, can we poor OSS types program against the PPU? Or do we have to drop 10k for a Novodex license?

      This would *suck* for the indy game industry.
  • Can I use it to do my quantum mechanics homework due in 30 minutes?
  • lamphrey (Score:3, Insightful)

    by achacha ( 139424 ) on Tuesday March 08, 2005 @04:10PM (#11880549) Homepage
    Unless they can make a deal with ATI or NVidia and have their PPU work with a GPU, it will be a very difficult thing to sell to people. They also need to get Direct X support and maybe have it work transparently with it (if that is possible). I can see this being a part of a video card, not a standalone PCI card unless the results are incredible and can be shown as a huge benefit to gaming, othewise only the hardcore framerate junkies will buy it.
  • It's fortunate that AGEIA didn't release this product on April 1st. I never would have believed it for one second.

  • Now thats using youre noodle, but it begs the question, is it a limp noodle an aldente noodle, of an uncooked noodle, all of which have different physical properties.

    Does a bug in the PPU mean that in fact a Eorpean swallow COULD carry a cocunut from point A to point B , when everyone knows it would require an African swallow ?

    In jest though would a bug in the PPU create a situation that could not be escaped in programming a game/simulation based on its "Laws" ?
  • I would be far more excited if they had any information on how this great new chip was going to be distributed. Are they working with ATI or NVIDIA to integrate this new chip onto existing graphics cards? Might be the only way they will make use of the full bandwidth on the PCI-E slots ;) Or are they working on getting a new type of card out onto the market? And if so, have they talked to the motherboard manufacturers on the type of requirements that it might need? Nice announcement. And I hope that it
  • I didn't RTFA (this is slashdot, what did you expect?) but what I'm wondering is how this is different from a CPU being dedicated to running the physics engine? Is this chip somehow optimized for physics? Right now, I can't think how it could be better than, say, having a box with 2 64bit CPUs and dedicating one of those CPUs to physics calculations when playing the game. To me, physics calculations should just be a bunch of relatively normal calculations (probably performed on a matrix) which just about
  • If we ever want truly virtual reality as we've all dreamed of it, we WILL need something like this. Remember, software can only do so much, no matter how efficient it is. I for one welcome our new PPU overlords since it will let us take a big step forward in letting games react like reality. Halflife 2 showed us how cool physics can make a game when done properly, and I look forward to more games doing that.

  • This reminds me of the Tom and Jerry chips from the Sega Saturn. One chip could great graphics while the other did great music. Alone they couldn't do much of anything.

    So how does this relate to this new chip? Simple: This new chip will do great physics processing, but with low-end graphics. I could see nVidia or someone else buy them out and stick the PhysiX chip into one of their card along with (or inside of) their current vector anti-alice-the-maid powerhouses. This would create a dual architectur
  • by X ( 1235 )
    You know, this kind of screams of snake oil, and I'm kind of surprised Slashdot posted it. As far as I can tell, there is no technical information on the actual product.

    I grabbed the whitepaper, and was disappointed to see nothing about the "PPU's" design. Here's the thing: managing a physics model basically involves a lot of floating point math, which CPU's do fairly well already. It can be parallelized a fair bit, if you know what you're doing, so you could build a processor which can execute more floati
  • They'll be coming out with a revision in a couple of months to simulate Star Trek physics. It will basically let you change the laws of physics at any point in a game, and will also spit out some technobabble explaining how you did it. (Reversing the flow of the barrion reverter, Captain!)
  • I'm curious about this. Admittedly I'm no expert in either graphics or physics, especially where processors are concerned, but doesn't it make sense to have both combined instead of separate?

    Now I don't know how a physics processor works but I'm assuming it's something like the following. You fire bullet (x) and player's head (y). The game feeds the PPU all the physical properties of the objects in question and the PPU figures how the bullet interacts with the player's head (snaps back, tugs the body
  • by Stiletto ( 12066 ) on Tuesday March 08, 2005 @04:30PM (#11880797)

    Reading the results of the "PPU" is going to be the stumbling block. Graphics accelerators work because you compose the geometry then send it off to the other processor, and from then on you don't worry about the data. You don't have to worry about reading anything back.

    I assume having a seperate "physics processor" will mean the app has to send the data off to be processed (say, a couple thousand points to collision-detect against a couple thousand planes), but then your app needs to read the results back across the bus! Is the time saved off-loading these computations going to be worth all this IO?
  • PCI-E bandwidth (Score:3, Interesting)

    by Anita Coney ( 648748 ) on Tuesday March 08, 2005 @04:44PM (#11880955) Homepage
    Doesn't a PCI-E slot have enough bandwidth to include a PPU on the GPU?! If a PPU is a good idea, then it would seem that nVidia or ATI could simply slap a chip on their own cards and sell them for more money.

    BTW, never in my life did I ever think I'd say the phrase, "PPU on the GPU"!
  • by Animats ( 122034 ) on Tuesday March 08, 2005 @10:52PM (#11884611) Homepage
    As someone who's written a serious physics engine [animats.com], I look forward to seeing this tomorrow at GDC.

    There's no one key item that bottlenecks a rigid-body physics engine, and it's not a simple pipelining problem like graphics, so the main thing special-purpose hardware can provide there is parallelism. And plenty of double-precision floating point power. (In a single-precision system, you have to take great care to never try to do physics far from the origin.)

    Collision detection is a minor CPU load if you do it right. If collision detection is using more than 10% of your physics time, you're doing it wrong. This may seem counterintutive, but the good algorithms are incredibly fast, even in complex environments. It's more of a data structure issue.

    Deformation, i.e. finite element analysis, is more of an inner loop crunch problem than rigid body physics. Finite element analysis has been parallelized for decades in engineering applications, and the problem is well understood. It's localized; you can divide the problem up into cells. So I'll bet that's what they are focusing on.

One good suit is worth a thousand resumes.

Working...