Slashdot Mirror
Chip Promises AI Performance in Games
Heartless Gamer writes to mention an Ars Technica article about a dedicated processor for AI performance in games. The product, from a company called AIseek, seeks to do for NPC performance what the PhysX processor does for in-game physics. From the article: "AIseek will offer an SDK for developers that will enable their titles to take advantage of the Intia AI accelerator. According to the company, Intia works by accelerating low-level AI tasks up to 200 times compared to a CPU doing the work on its own. With the acceleration, NPCs will be better at tasks like terrain analysis, line-of-sight sensory simulation, path finding, and even simple movement. In fact, AIseek guarantees that with its coprocessor NPCs will always be able to find the optimal path in any title using the processor." Is this the 'way of the future' for PC titles? Will games powered by specific pieces of hardware become the norm?
sounds like it just speeds up existing AI routines..... and existing AI routines, well, SUCK.
I dont think we are going to get any good AI until it has some method of "learning"
Who is Al and why do I want him controlling everything in my games?
This guy's the limit!
What may occur is a separate box consisting of the GFX card, Physics Card, AI card, PSU for the above along with supporting memory modules just to power existing games. Mulitple cards consisting of mulitple chips with multiple cores will likely overpower the common case. Thus for the hardcore games, a separate box wired to the main rig could be the norm. Thus, for the average home user, we will get smaller and smaller (Mac mini et. al) but for the gamer we'll see module system, with multiple boxes and multiple PSU's to help with cooling and overall performance goodness.
Well if Chip promises it, I believe him..
The physics card could theoretically work because if the player doesn't have it, you could always leave out some of the eye candy and only calculate fancy physics for objects that affect gameplay. With an AI card, you don't have that luxury. Either they player has it, or you have to just dump all the AI (obviously not) or do it all on the CPU, which begs the question: why program your game for a dedicated AI card if you're just going to have to make it work on computers without one?
Against stupidity the Gods themselves contend in vain.
Something that's always bugged me a bit about expansion boards is that the experience can only be enjoyed by the user with the board.
For instance, in a multiplayer game, some players will obviously be getting better graphics than the rest - but often the maps are tailored to work equally well (or at least as equally as possible) to low-end and high-end video cards.
And then there is this new physX card - which sounds like a neat idea, but you have the same kind of situation. You can kind of model physics looking a bit better for the player with the card - but all actual physical actions must be reproducible for the non-card having players.
Now, here is where I think the AI card could be different: distributed processing.
Let's take two human players and 4 AI players in a multiplayer game. Normally the server would be responsible for the AI decision-making processing and would pass to the clients only the x,y,z movement and animation data as a network stream. The AI thinking would take place completely free of the client machines. This puts strain on the server's resources.
Now, imagine rather than the server processing and the clients recieving network info you were to turn this on it's head.
Have the clients process a subset of the AI - say, 2 AI for player 1's machine, and 2 AI for player 2's machine. Now both clients will send the AI's movement information to the server. From the server's point of view the AI would require the same processing power that a regular human player would require (very little - relatively speaking).
With the plethora of bandwidth available client-side these days I think this kind of idea is very realistic.
My Computer Music Tutorial Videos
Aren't many problems of that ilk NP-complete?
They want to completely ruin game performance by killing the PCI bus bandwidth and causing the GPU to stall waiting on the position/orientation and generated geometry that it will have to render?
Physics and AI coprocessors are 2 years too late - with the increasing availability of dual core processors in even midrange consumer systems now, and quad core on the horizon, engineering time is much better spent on making an app multithreaded so that it runs efficiently on hyperthreaded and dual core machines, instead of trying to offload it to a coprocessor that few customers will have. For a consumer, it is a better investment to spend an extra $50 to $100 for a dual core processor than spend $300 on a physics or AI coprocessor.
I doubt, and openly mock, their claims of '200x' speedup. I imagine it will be more like speeding up the process of $200 leaving foolish consumers' wallets.
Hunt your preferred prey at Aliens vs Predator MUD. Join the war at avpmud.com port 4000
I find this interesting, but i have to agree with above posters. My line of thought is when it comes to multiplayer games the server needs to be doing things like physics calculations. The server should also do AI. I wonder when we will see the first MMO with a realistic physics model. I want to see a MMO that starts up AI to go about the universe when their are very few players on, or to help the game be fun when it is starting out and their aren't that many people on.
You mad
If things continue in this direction, it looks like we may be buying game consoles to hook to our computers instead of our televisions.
Slashdot Burying Stories About Slashdot Media Owned
The AI card won't work for the same reason the physics card won't work. I'm not buying another card. PC gamers already need to spend an extra $200-500 for a graphics card in addition to the rest of the computer. I'm fine with that, but like hell if I'm buying 2 other cards for other features. The only chance these things have is if they get it all onto one card.
The product, from a company called AIseek, seeks to do for NPC performance what the PhysX processor does for in-game physics.
Damn. And I hoped it'd actually be useful for AI.
The problem with PhysX is it costs similarly to a mid range graphics card and yet adds kind of performance gains of a first gen graphics card. Whilst GPUs are massively evolved compared to first gen offerings, PhysX in its first gen state is a really expensive nice little add on.
Don't get me wrong, when physics processors and AI processors make the kind of difference that having or not having a GPU makes now, that'll be an amazing thing. It's just that, in first gen form, with no competition yet pushing the market forward, PhysX has been met with a deafening yawn and saying this chip hopes to do the same doesn't really promise much.
Seems to me that I recall these exact questions being asked about 3D accelerator cards a number of years ago. Why program for it if you're going to have to make it work without it anyway? For a few games, they just simply made the game not run if you didn't have the particular piece of hardware (this memory is tied directly to a version of a BattleTech-based 3D game that relied directly on you having a Matrox Mystique card, though I read about other things which relied on various Voodoo or... what was it... eh, it was another discrete 3D card that didn't even interface with the normal graphics card. I had one at one point. It was neat when it worked and sort-of supported Glide but with interesting distortions and broken-ness-es...
-----------------------------------------
Remove the Greed which plagues mankind.
"Heartless Gamer writes to mention an Ars Technica article about a dedicated processor for AI performance in games. The product, from a company called AIseek, seeks to do for NPC performance what the PhysX processor does for in-game physics."
Or both can be used to speed up Vista.
Since the Mhz jumps of the past seem to be by and large behind us these days, but we're looking at more and more cores, isn't it time that games become multithreaded and offload that nasty pathing work to a second core? Sure you could buy stupid shiny cards for the game physics and AI and network (some sort of network booster that avoids the OS's TCP stack -- posted a while back I believe), or alternatly just make use of the extra hardware that /will/ be in the box anyway.
Now, the decent AI toolkit that folks can license might be worth it anyway, when they figure out they should just run it on the CPU instead of their custom CPU-like-thing.
Slashdot Patriotism: We Support our Dupes!
I think that AI and physics co-processors have a better future as part of the CPU, rather than an add-in board. Perhaps as additional core(s) on an AMD processor, with full access and feedback to the CPU proper?
The cesspool just got a check and balance.
The first is money. A serious gamer who likes his bells and whistles might be expected to spend several hundred dollars every year or two, in order to make his games run at their prettiest and fastest. He still has a finite budget, though-- asking him to spend a similar amount on physics and AI hardware is unlikely to have the desired effect.
The second is developer support. Developers are stuck in an even bigger pickle: on the one hand, these devices (ideally and theoretically) provide new avenues for gameplay, but the moment that the hardware becomes necessary, they've eliminated a definite percentage of their market.
Three... are these things necessary, or even desirable? The original PhysX demo application, intended to show the effectiveness of the hardware by flinging crates around, ran perfectly smoothly on good hardware once hacked to remove the check for the PhysX processor. The Killer(TM) NIC is pretty much marketing snake-oil to anyone with any knowledge of networking. The 'need' for an AI coprocessor is pretty much obviated by faster main processors. Most games these days haven't been optimized for the multi-core processors that, unlike parlor tricks like PhysX, are actually growing in popularity. Wouldn't it be just that much easier for a developer to assign AI routines and meaningful physics interactions to idle processor cores, rather than constantly shuffling vital data back and forth between peripheral cards?
As always, computer "innovation" goes in circles. Everything goes from dedicated hardware to software running on a general-use CPU, and back again. Terminals are replaced by PCs, which become hosts for remote interfaces to applications running on servers somewhere else.
It seems like everything I can do now I could do in 1990, but today I can do it *in hi-definition, wirelessly*. Yippee.
Insisting on "correct" English is like saying that there is only one, definitive recipe for chili.
Slap this into the Xbox 720 or PS3/4 and you get a mondo increase in NPC performance as long as the developers put in some rudimentary "learning" routines to keep things random. All gamers buying games that desire that NPC chip, get to enjoy the fun. Not so for the PC gamers. For PC gamers, game companies that make games for the most elite configurations - namely, those requiring the PhysX processor and this one - will have a lower percentage of sales per owners of PCs.
--- Grow a pair, liberals... stop letting the Republicans bully you!
I could see something like this used to lower the costs, and increase the scale of games like Everquest/World of Warcraft. Those games have dedicated server machines running AI's 24/7, for profit.
"sounds like it just speeds up existing AI routines..... and existing AI routines, well, SUCK."
So much for "chipping" slashdot posters.
Is this the 'way of the future' for PC titles? Will games powered by specific pieces of hardware become the norm?
Short-run, maybe; long-run, no. IMHO, things will consolidate like they always seem to do. Video cards are necessary for more than just games, so they won't be going anywhere. Physics and AI cards seem to be useful for nothing but games. It would be foolish to combine all video cards with physics and AI chips because not everyone plays games, but why not combine the physics and AI chips? Farther down the road someone will come out with a new card to enhance some other aspect of gameplay, and eventually that will merge with the physics and AI chips on their own card.
Things are always being consolidated on PCs. Look at all the things on mobos that used to require separate cards 10 years or even 5 years ago. Designers get better and better at cramming more things into a smaller space (even if that is getting harder and harder to do), so it seems to me that these things will keep merging together when it is useful to do so. In this case, I don't think most PC users want to have 3-5 cards just for games, so it is useful. I could be completely wrong on that point though.
> Is this the 'way of the future' for PC titles?
I mean, I care, but best-selling games has always been about licenses, tie-ins, snazzy graphics. If there's one thing I've learnt from playing (and to a lesser extent, writing) computer games is that nobody (statistically speaking) cares about gameplay and AI.
Besides, the best games I've played recently - ie the Battlefield series of games from EA - don't even use AI (unless you're in billy-no-mates single-player mode).
One begins to wonder what the "endgame" scenario for the respective manufacturers of the physics and AI cards we're seeing. I can foresee three distinct situations:
1) The CEOs, investors, and engineers are complete idiots, and expect all the gamers of the world to buy separate physics, AI, and graphics cards
2) They're hoping to provide chips to ATI or nVidia for a "game card" instead of a "graphics card", the next generation of expensive purchases for gamers
3) They're hoping to provide chips for the nextgen xbox / playstation / wii, hoping that their chips will be the ones to make gaming interesting again.
How am I supposed to pretend to have quad SLI, PhysX, this AI card, Soundblaster X-Fi, and the $279 'killer' NIC all at once, huh? It's just not fair!
This is a great concept, but all the same, more imposing on the Game dev's and hardware vendors. In my mind this is just another API that you would need to study the libraries for, and would consequently drive up the development time for a project; that in this day and age will slowly kill of hype before release, and then kill off SALES when the end-user discovers that the shiny new AI processor hinders the bliss of that which is 60+FPS... If you look at the PhysX cards and all the hype that it generated as an example, this kind of 3rd implementation actually makes the games perform sub-par. Also, I have not seen any tangible difference in-game between physics performance with PhysX and without PhysX; you will only see graphic's performance which does not equal "physics performance". So until the chipset manufacturers(Intel, Nvidia, Via) get involved with the providing solutions to allow these products do what they are advertised to do, they are all for not.
As for this new thing "doing the same thing as the PhysX processor", we'd have to see this PhysX processor in action (and on the market) first, wouldn't we?
I assume you could accelerate A* with a dedicated chip - but that makes only a relatively small part of AI. Or you could accelerate Neuronal Networks, but most games I know use relatively plain state machines.
I'd move the pathfinding onto another thread, and with the gaining popularity of multi-core architectures you should get the same effect. That way you'd share most of the resources with the rest of the system, and wouldn't have to worry about sending everything over the bus to another card.
I agree. Intel just released dual core chips, AMD has them already and is about to release quad core chips, plus we have -cheap- dual processor boards available. That'd be eight cores, as soon as AMD releases their new kit.
Even Windows is shipping with SMP available, we have processing capability out the wazoo pretty much. Should be able to handle any AI requirements I'd think and have room to balance your checkbook at the same time.
Some clever lad should be able to design a bot that doesn't do the same thing every single time, eh? Maybe learns to check that blind spot before it sticks its head out. Now THAT would be fun! Better than new eye candy, for sure.
True, finding the optimal path is often NP-complete, but if the AI card decreases the constant factor in the majority of cases, this could still be a win.
They want to completely ruin game performance by killing the PCI bus bandwidth
Positional updates to a character in the game are very low bandwidth - I mean, MMO's do this all the time and don't saturate network connections, much less a PCI buss. The calculations are heavy but the input and end result are just a few numbers, plus a terrain map you would load once and forget until you zone, at which time a little latency is happening anyways.
causing the GPU to stall waiting on the position/orientation and generated geometry that it will have to render?
Read carefully. It isn't generating terrain, just sending around updates. Diffs between 2 meshes don't have to be big. The mesh will probably stay the same, just relocate. Send an array of updated points with the corresponding indices. It isn't hard to imagine that a dedicated processor could do these things significantly faster than a processor that is already breaking its behind doing thousands of matrix transformations, player calculations, sound and graphics effects, etc.
Perhaps the card could be most useful not on the client, but in dedicated mmorpg servers. I know WoW could definitely use some smarter mobiles. Sometimes I think whoever designed the AI was inspired by the green turtles from Super Mario 1. I'd like to see games with smarter mobs and NPCs, and any game with a realistic ecology (for instance, suppose mobs don't magically spawn, they procreate the old fashioned way, and must eat food (a limited resource) to survive) would require many more mobs than a WoW-like game in order to prevent players from destroying the environment. Simulating millions of intelligent mobs would likely be very expensive computationally.
I've been using nvidia forever and a day since I switched out from the Matrox Parhelia.
When is Matrox coming out with something new?
--- Grow a pair, liberals... stop letting the Republicans bully you!
So far, those seem really short of AI. Maybe because they have less computing power per player (that server farm must be affordable). With dedicated AI cards for the servers, MMORPGs might be able to catch up to newer single player games that have at least half-decent AI.
C - the footgun of programming languages
Will video game consoles for computers come with the same systematic bias against smaller game developers that video game consoles for televisions have traditionally come with?
Ok I do know I should be more tolerant of my fellow man and all that stuff, but really... this is just damned foolish.
Imagine the conversation that led to this...
-misty flashback fade-
Marketing Guy : Oh man, gaming is ready for a revolution!
Technical Guy : It's called a Wii now
Marketing Guy : Huh? We now what? -shakes head- I mean these gamers, they buy top end stuff, they have money to burn!
Technical Guy : Not really, they buy slightly under the curve and tweak up and overclock mostly
Marketing Guy : No no I read in a magazine that all gamers have more common sense than money
Technical Guy : -sigh-
Marketing Guy : These Ageis guys really whipped up a lot of frenzy about a new type of add on card.
Technical Guy : Yeah it's supposed to make the gamers run better by adding physics processing but the demo..
Marketing Guy : And they are making money hand over fist!
Technical Guy : Well, actually...
Marketing Guy : And it's so easy to make specialty stuff!!
Technical Guy : But their demo runs the same even without the card!
Marketing Guy : Wait, Wait, I got it! We'll make a card that adds more CPU power!
Technical Guy : Well dual cores add lots of CPu power that has yet to be tapped by games
Marketing Guy : No wait, even better, we'll make it special! That's what made the Ageis guys rich!
Technical Guy : Listen, the Ageis guys are not selling much, you might not want to...
Marketing Guy : We'll add better AI! That's IT!
Technical Guy : Better AI?
Marketing Guy : Yeah, we'll sell a card that makes the games run better!
Technical Guy : How's that work?
Marketing Guy : We'll umm, make it able to process AI commands like a graphics card processes graphics commands.
Technical Guy : But Graphics Commands are standardized, so they can optimize for that.
Marketing Guy : We'll get them to standardize AI commands.
Technical Guy : -twitches- But, every game has different needs from AI
Marketing Guy : So we'll make it flexible, generic, so it can do anything
Technical Guy : If it's generic processor design, it's the same as a regular CPU.
Marketing Guy : Exactly!
Technical Guy : But then what is it's advantage?
Marketing Guy : Haven't you been listening? It'll make games play BETTER!
Warning: Teh poster of this messaeg is lysdexic
...that's I've always wanted the answer to from someone who knows what they're talking about:
For the application you've described, and similar ones, people always claim it would be cool to be able to handle massive dataprocessing so you could have lots of AI's, and that would get realistic results. However, it seems that with *that many* in-game entities, you could have gotten essentially the same results with a cheap random generator with statistic modifiers. How is a user going to be able to discern "there are lots of Species X here because they 'observed' the plentiful food and came and reproduced" from "there are lots of Species X here because the random generator applied a greater multiple due to more favorable conditions"?
I saw this in the game Republic: the Revolution (or was it Revolution: the Republic?). It bragged about having lots and lots of AI's in it, but in the game, voter support in each district appeared *as if* it were determined by the inputs that are supposed to affect it, with a little randomness thrown in. The AI's just seemed to eat up cycles.
Long story short, aren't emergent results of a large number of individual AI's essentially the same that you would get from statistical random generation?
Apology to Ubuntu forum.
It seems that those guys did what's best under these circumstances - got a specific search space that is common in many games and specialized in that. IMHO, it's not enough to get the snowball rolling, but time will tell.
The obvious application for this technology is in MMOG servers, not in desktop machines.
I didn't know what the Killer(TM) NIC was, so I looked it up. It's $279 for a NIC. That's all anyone need know. They throw some buzz words out there, but remember, $279. It's a space heater/NIC that takes up a PCI slot and costs $279. End of story.
$279
> In fact, AIseek guarantees that with its coprocessor NPCs
> will always be able to find the optimal path in any title using the processor.
It has been mathematically demonstrated there is no general pathfinding solution significantly better than trying all possibilities (though pretty much only in degenerate cases could the best path be difficult not to find by a hill-climbing heuristic.)
Still, it should be trivial to whip up a case that would require these dedicated processors longer than the known age of the universe to find the optimal path.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
You mean a separate box just for games? What would this mysterious "play station", look like, what would it cost? I'm having trouble imagining what this "game cube" or "entertainment system" would cost, and who would buy it. Who would write games for it? It's an unknown, or an "X box", if you will.
Nah, it will never happen. People like to use their hardware for word processing and spreadsheets, not shooting aliens. These aren't "game boys" we're talking about here.
Now if you'll excuse me, I have to go Wii.
This will be a big win for Hyper Transport based co coprocessors / cards if amd can make real easy for chips like this to have a hyper transport bus and would force intel to use Hyper Transport.
By the time intel gets to there own hyper transport like bus there may be a lot of hyper transport cards / coprocessors out there and It would be a vear bad move for intel to try to push there own csi bus.
You've done a very good job at being condescending so people mod you 'insightful', but you don't really know what you are talking about.
There is no way that a generic CPU, even four of them, could compete with a custom piece of hardware. If you don't believe this, try running any first person shooter on a dual core machine with your graphics card drivers uninstalled. It will look like a piece of crap.
In my digital design laboratory we made a video processing unit on a 100 MHz FPGA and compared its performance to a 650 MHz CPU. The FPGA was hundreds of times faster, and we weren't even doing anything clever. These custom AI/Physics cards have many cores with parallel computing power and insane memory bandwidth so they can pump many, many more computations through than a generic CPU.
Of course, the 'average user' would get more mileage out of an extra core instead of this AI card. But they were never intending to market this to the average user. Personally, I think this will have a very hard time catching on for all the reasons other people have already stated.
On games I've worked on in the past, we had a global strategizing algorithm that ran once every few seconds (over the course of a bunch of frames), more localized map sectional AI that ran slightly more frequently, per-unit pathfinding that ran (incompletely) every second, and moment-to-moment movement that ran every frame.
Now, if we could run all of those AI routines every frame, the game would appear a bit smarter. It wouldn't have a delay upon reacting to stimulous, the pathfinding could run a character intelligently across the map without bumping into dead ends, New units would path immediately instead of waiting for the next global strategy cycle, etc.
Not a major update, but a perfectly scalable one.
The ______ Agenda
Hmm, I thought we had that already... What about all those game consoles with custom video chips and CPU's in them? (PS1, PS2, Xbox, Xbox 360, Gamecube...)
IMO, this chipset (or at least its functionality) may be more likely to find a home in consoles than as add-on's for PC systems.
As others have pointed out, the number of people who would drop an extra $100 to get the last erg of performance out of their gaming system is pretty small.
It's real; They already have a chip manufactured which they demonstrated in recent shows.
They have a playable game that looks just like one of the videos on the site (The one with the soldiers; although the game graphics looked a bit different when I played it), in which you control the Tank, and the soldiers act intelligently.
Destroying the city is extremely fun, and it was the first time ever that a large group of AI-controlled agents looked intelligent to me.
Hope they make it to the market soon. If game studios use this coprocessor properly, games are going to become much realistic, and much less micromanaged.
... a chip that plays the games for you, so you dont have to!
You never needed a serious firepower for a computer opponent that pretends to be actually doing something before ultimately rolling over. To think that humans can even compete against computer in games that tests physical accuracy (computer is infinite more accurate than any human) or multitasking is simply foolish.
AndrAIa is my favorite AI. Making her 200X better is an amazing throught to speculate on.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Graphics accelerator, physics accelerator. Now we have AI accelerators. I don't to have to buy multiple accelerator cards next time I buy a computer to play the latest game. What we need is some sort of generic chip that does all of this. Oh wait...
We have dedicated 3d ASICs on our graphics boards now, we're considering physics ASICs, now AI ASICs? It's time to end this silliness and start shipping FPGA's in consumer products!
Here is a DIY plan for very good AI:
1. Purchase separate PC or console system
2. Purchase steel cage
3. Kidnap gamer from a nearby neighborhood
4. Put gamer in steel cage with game system
5. Feed gamer only when he complies
6. Write API to take advantage of gamer prisoner
Profit!
I really enjoy letting the marines in Halo 2 do the driving because they kind of understand that it's bad to get near cliffs, but they kind of don't understand it as well. The MC does the driving from here on out.
Can I bum a sig?
This touts the ability to speed up AI 'thinking' or 'reaction' times. Is this something that really needs to be sped up?
I believe the current situation of AI is a result of lack of breakthroughs or lazy programming, not that the AI simply reacts too slowly as a result of your CPU not being able to process it fast enough.
If this is truely the way the market is heading though, and not just a hardware bubble, then I think we will see things like this integrated onto high performance motherboards rather than as a seperate device.
I always joked with some hardcore pc gamers that you needed the NASA computers to play nowadays PC games. Looks like it will soon become a reality.
So for optimal performance, I need two video cards, a physics card, an AI card, a sound card, and a network card. And even then, that's leaving out stuff like a RAID or SCSI controller. Sounds great, but where's a motherboard that can support more than one PCI card with both PCI-E slots filled? Hell, a lot of motherboards can't even handle one.
In Soviet Russia, backwards is everything.
This is just another spoke in the wheel of reincarnation. This too shall pass.
This space for rent.
A lot of you seem to think that neither this AI card, nor the Physics card are going to be of any use, but I don't see it that way.
The physics card for example, is really just a processor dedicated to vector processing. Couldn't you use this to simulate light bounces? Then you might be able to up the graphics further with a sort of partial ray tracing. Additionally, this'll help with intersection detection, so if you swang your sword, it wouldn't hurt the other guy until it actually intersected with him. It may also have implications for software based synthesizers... rendering audio is much more intensive than video, perhaps a physics card will be able to help. How about water that ebbs and flows realistically? Clothe caps and realistic hair? Besides, having largely mutable worlds in video games is just sweet. A second processor will definitely help in this regard, but a CPU dedicated only to vectors will very quickly outpace a single additional core in only a few generations.
The AI card on the other hand is similar, except its optimized for tree searches. This could be used in a number of applications -- chess and Go for example. But also perhaps for more fuzzy strategy, like RTS games. This won't only help with moving an entity in a 3d world, perhaps it can also help those entity's think better, by allowing them to consider more alternatives before making a decision.
Certainly you can do all this on a dual core system 'because its fast enough' but then why not turn down the graphics quality on your games, get a quad core machine and run it through an OPENGL Emulator? If these catch on, they'll eventually out pace generic CPUs for speed in their purpose, and the games will use them. Not having hardware AI/physics support then will be like not having hardware graphics support now.
That will drastically alter gameplay. You'd literally have to design the game twice, once for dumb AI, once for smart AI. As an example, look at the difference between the original Doom, and Doom 3. While Doom 3's monsters aren't brain surgeons, they are smart enough to sneak around, take cover, etc. If you were to apply those tactics to the massive numbers of monsters in the original Doom, you'd have a near impossible game. Likewise if you put the dumb, "walk straight at the player" AI in Doom 3, the challenge would be gone.
Now this is just the case with a game where AI is fairly unimportant in the scheme of things. In a game where it highly relies on the AI, say one where squad tactics are used, it'd be a nightmare. With the card you have highly competent teammates that practically complete a mission for you, without it you have guys stepping on their own grenades, things like that.
AI also has the problem of being different for different games. I'm sure the AI process for an imp in Doom 3 is nothing like the AI process for an enemy civ in Civilization 4. Thus I don't know there's a way you can provide a more "optimised" kind of chip for it. Graphics accelerators work because you can design a chip that's highly specialized. They'd suck as CPUs, and in fact until very very recently weren't even Turing complete. However since graphics is always the same kind of thing, they can be optimised to do certain things very fast. I just don't think that's the case with AI, since there's so many kind of AIs one might need.
entertain this notion? This card is doomed.
The PhysX card is doomed
Multicore CPUs are here - no longer some weird expensive ninja-component, they merely cost a few $/£ more than a single core.
Currently nothing (non-industrial) really takes advantage of multi-core systems - the spiel for them currently seems to be 'Run an AV scan without slowing your game' - that's it.
*rubs crystal ball*
What's going to happen is the established middle-ware (i.e those with a product people use now) will develop engines that 'run on a core'. Current core #1 will run the game and core #2 will run the physics and eventually core #3 will be the AI, #4 will run the procedural graphics, #5 will do the 12.1 audio etc.
If you look what's being developed for the PS3 (the most insanely multicored CPU so far), the cores are being divided up by function - one for the OS menu, one for the lead characters hair etc. Threading a single function across multiple cores is not only insanely hard, but hinders cross-platform porting.
Middleware is just going to be sold to run on one core and ported per platform - and I'm fine with that.
I think that once PC games start making us of these Physics accelerating hardware, and the AI accelerating hardware, I will be done with PC gaming. This is why consoles work so well. I never have to upgrade them, accept maybe with a few accessories, and developing for them is even easier for them because there is only one platform to develop for. Not to mention, this new hardware can be brought to the next generation consoles, while on the PC I would have to go out, buy a motherboard with the appropriate number of slots on it to fit a bajillion cards, and chips, just to keep up with the maximum of 5 good games coming out every year, or face playing them in some crippled, bear-bones mode.
As it is, I only buy a few games a year and I consider myself a relatively serious gamer. I even have a high-end nVidia GeForce card in my gaming PC. Still, the card was a gift that I would not have bought for myself, and I really have no plans on upgrading it for at least a few years because PC gaming is just way too expensive considering the number of quality titles coming out for the PC, which has only been going down recently.
I mean, what does a high-end nVidia/ATi (AMD...) card, new flavor of the month chip (AMD 64, whatever), Physics chip/card (I assume it goes into an AGP slot or even a precious PCi Express slot? Or do I have to buy a new motherboard with a chip slot), and AI chip/card (same question) going to cost? And why would you buy that instead of a console, which gets all but most strategy games (the reason I am holding onto the PC gaming) these days.
Oh, the new Far Cry game claims that it is too powerful for the Xbox 360, or maybe it was even the whole generation, but really, the 360/PS3 are pretty beastly, and it sounds more like a lack of interest to port the game engine than anything else. Otherwise, my PC has no chance and as I said, it's in pretty good shape.
If this does for AI what "3D accelerators" did for graphics, then AI is doomed to atrophy.
Prior to the invention of the 3D accelerator card, 3D graphics was awash with variety and innovation: Duke Nukem 3D's sprite-based engine allowed 3D Realms to simulate real-time mirrored surfaces; Shattered Steel used voxels to create a smooth, contoured landscape oozing atmosphere, then dotted it with metallic polygonal buildings and polygonal enemy vehicles; the Wing Commander games of the time used phong shading to give a strong metallic sheen and mimic the harsh directional light of space; while Earth-based flight sims stuck to full polygon engines used the computationally less intensive gourad shading to handle the higher polygon count of terrain modelling, also putting a softer shade on the polygons; and some people were even exploring the possibilities of early ray-tracing technologies.
Each game had its own individual visual style.
However, now every game generates a polygon landscape, with polygon enemies and polygon allies, all shaded in a soft, plasticy gourad shading. All alike, all indistinct; no style, no individuality; the unique strengths and weaknesses of different styles of rendering all swept aside in favour of a "convenient" gourad-shaded-polygon model supported by cheap consumer electronics.
As a result, improvements in 3D graphics technology have meant little more to the gamer than an increased polygon count.
By hard-coding AI routines into silicon, AI development will be similarly stunted, with innovation seen as the overpriced alternative to relying on what's on the accelerator board. The end result for the user is a more predictable, less interesting experience.
HAL.
Got them moderator blues I blieve I walk out the do', With these mod-points I been gettin', I 'most never post no mo'
why don't they make a card or 2 to add on modules that do graphics, ai, etc. just plug into a pci card, instead of 200000 cards to do 1 thing.
If your neighbours roof is flying past your window, you know it's cyclone season.
Program the physics, graphics and AI routines into hardware. Offload the processing onto the FPGA. Call it a generic games accelerator. The games developers could then optimise their own libraries of hardware routines for their games rather than trying to optimise the games for general purpose hardware.
Deleted
so basically in a year we won't even need system processors and system memory to play games?
instead we'll just need 203954 PCI slots?
seriously, this fad needs to end.
GRAPHICS CARDS! SOUND CARDS! PHYSICS CARDS! NETWORK INTERFACE CARDS! AI CARDS! what's next? honestly, this it just stupid. there's a border between reason and obsurdity.
AI cards seem to break that border. also I think physics cards should just be bundled with GPUs on a single card.
and network interface cards? might be neat if they cost around $30 instead of say... 300 friggin' USD. or if it was just another part added into motherboard shopping (a somewhat comparable example to what I mean can be found in raid)
From a technical point of view it sounds more of the ways things have always been. New hardware architectures are introduced to problems for which CPUs are not designed for and are outperformed. After few years they either become part of your CPU (FPUs) or your motherboard.
We met with the AIseek guys in GDC. If you remove all the marketing speech they have built a piece of hardware dedicated for solving problems on graphs. Useful in many fields and also in many problems in AI, and specifically in game AI.
What do you think HyperTransport is all about? Bye-bye FSB, hello massive amounts of bandwidth and direct communication between processors! We could theoretically stack EVERYTHING (network, cpu, gpu, spu, USB/FW/802.xx/BT, IDE/SATA, memory controller, all of that and more!) on one die and have a computer the width and depth of a mac mini, but half of it's height. I say hell yes to that.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
I read a year or so back about a computer that used a "bookend" approach to being built. You started with a main vertical unit, then you have universal connectors on either side, in the shape of a circle. You could simply slide out the locking mechanism, add in another module, lock it back, and boom, instant upgrade.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
I went to the AIseek website to find more information about the architecture. I found the whitepaper, and while they refer to the technology as a 'processor', I think it's more in the sense of a 'software engine' rather than physical hardware. Nowhere do they mention any architecture or implementation details, yet they compare and contrast existing AI algorithms to theirs. What do you think? Is there really a chip? -Phil
Graphics and sound cards work well because they are almost purely output devices: mostly, you throw stuff at them and forget about it. Physics and AI are another matter entirely: very tightly coupled to game code. Round-trip latency even on a fast bus like PCI-E will kill you on such fine-grained calculations. Reports I've heard of PhysX's performance bear this out: up to an order of magnitude SLOWER than CPU-based Havok when performing real-world (OK, real-game) physics. I don't expect this device to do any better for AI.
AI Programmer here.
;)
The computational requirements break down into several areas:
Pathfinding:
Finding *a* path isn't the problem. Finding an optimal path is a lot harder. This is especially the case as you add realism, such as varied terrain that impedes movement in different amounts.
Also, consider games that deal with nonstandard paths. I worked on Alien vs. Predator 2 -- try implementing an efficient pathfinding algorithm for Aliens who can crawl on walls and ceilings. Oodles of fun
I'd be willing to bet that there are many games with cases that are not solvable in polynomial time and still require a heuristic even with a hardware AI processor.
Sensory input:
A lot of this involves "can I see foo" or "can foo see me". LOS checks are usually expensive as you are either raycasting, or you're doing an intersect segment. If you want an exact distance, throw some square roots in there for good measure.
Movement:
This falls more under the physics category. From a pure AI side, the only thing special is the state machine running the AI's behavior(s).
Anyhow, take what you will from it.
Since the Mhz jumps of the past seem to be by and large behind us these days... MHz capital H Please show the man some respect. http://en.wikipedia.org/wiki/Heinrich_Rudolf_Hertz
In case you don't know who I am talking about.
Maybe it's just my pet peeve, but it's getting really tiresome to see every half-assed heuristic or simple algorithm described as "AI" just because it's used to control the action of some in-game object. This sounds like it might be a useful chip, but AI it is not.
Proud member of the Weirdo-American community.
First, graphics won't work on an FPGA. I mean technically it's feasible, but the demand for graphics is great enough to make it economical to produce graphic ASICs such as those Nvidia and ATI produce.
However the FPGA idea is a good one, and is being researched. Actually, what is even more interesting is to utilize transistors on the CPU die to integrate reconfigurable hardware accelerators. The research is being done currently, and will allow for CMP + reconfigurable systems so that custom processors can be integrated at runtime. Additionally these systems will allow dynamic reconfigurations, so that the hardware can be time-multiplexed between different configurations (although the configuration time can often be on the order of miliseconds). Additionally, this allows for straightforward upgrades to your processors. Want to run games faster? Throw more hardware, double the size of the reconfigurable area. This will allow the game to do more in parallel, or to cache configurations etc (causing more routines to run in hardware rather than software). And everything is compatible as the user just needs to download the synthesized version of the hardware for their reconfigurable arrays.
Of course this scenario is quite a ways in the future, and it must be remembered that hardware designs are much harder to create than software designs, and hardware/software design is a very difficult problem. The future is there, it just may take a while for reality to catch up with the research.
Phil
Although I agree with you, and think FPGAs are hot, they are also REALLY expensive for the models that game developers would use (they tend to write big software from my experience, so your cheap FPGA wont have enough gates, and scaling for different users would be a bitch) BUT, if you had a processor that had a memory of some sort, and regognised series of tasks that it did, it could program an FPGA to offload things that it recognises... which gets back to the need for at least a little AI to determine said tasks. I think it would be better fit for sitting next to the processor, instead of off to the side.
Your post reminded me of the game Outcast which is voxel based. The game is both ugly in a blurry pixelated way and beautiful in a geometric way. Unfortunately they limited the maximum resolution to the hardware capabilities at the time the game came out. Current hardware could play it at much higher resolution and it would still run smoothly and look a lot better. The game itself is not to be missed IMHO, though many will be turned off by the blurry graphics. If you can look past that you can see a real beauty to the voxel landscapes.
That's disappointing. When I saw the title, I was hoping it was talking about AI issues which actually make games impossible to make, like the ability to cater for player actions not anticipated by the programmers or have non-scripted conversations. Instead it's just part of the endless drive for "More Performance!"
Pretty much what I would have said.
"seeks to do for NPC performance what the PhysX processor does for in-game physics" You mean it plans to do absolutely nothing?
+1 IDisagreeSoHeMustBeATrollOrAnAstroturferOrAShill
They're mostly used in custom applications at the moment, relatively low production runs makes them expensive. If there were such a thing as a generic game accelerator card they could be produced by the million, the cost per card would come down to well, damned near nothing in the long run. I think we'd see a single design/model dominating much the way the ix86 architecture dominates just now. It takes 2 things, 1, one of the producers to package a decent fpga along with an easy to use development environment and some pre-built designs and 2, a killer game which couldn't be done without it.
The added benefit would be that eventually most machines would come with an FPGA built in, they would pretty quickly start to be used for other applications.
Deleted
Maybe they make it into a USB device, include it with their game and require you to have it plugged in to play. That way everyone who bought the game has it and who doesn't, well, doesn't play. It could happen if the AI chip was cheap enough to produce. So you have a half dozen titles, plug them all in and have them load share.
Sorry, I was busy rigging up my Aimbot card into that ISA slot.
Tell me something...it's still "We, the people"... right?
Um, computers have multiple cores. Unless I am on drugs, I don't think games are using the extra core(s) as much as they could, if at all. How about making one core in a dual core processor do the AI calculations, and the other core do the "main" stuff? Since cores are only going to increase, this seems logical to me. It's not like any kind of card you plug into your computer would ever be accepted by the market/developers.
You've got a friend in Japan: http://www.jlist.com
So will script kiddies be able to take advantage of this for cheat bots?
I can't see how this will be worth the money, the PPU was a joke. While this has it merrets Quad Core CPUs are just around the core'ner. At that point you have 1 core for game logic, 3 shared between gfx & AI. That's a whole lot of CPU cycles!
In grand prix legends this was different. You will loose until you get very very good. But somehow not winning in an extremely hard race with AI drivers who seem to be driving with the same ruleset as you is a lot more satisfying.
Imagine a war game. Wouldn't it be a refreshing change if for once you played the loosing side. If you had to retreat because of the AI and not just because the script told you too? That Wing Commander mission where you loose your wingmate would have been a lot better if the AI had actually been capable of beating you, or even being a challenge. Instead you knock them done left and right and still loose because that is what the script says happens.
I rather loose in chess against kasparov then win against a retard.
Also I think you mistake good AI for unbeatable AI. With good AI you can made the AI beatable by being smarter. But if the AI is dumb it has to rely on its units being overpowered to make up for their weakness. Hence why 99% of console racing games always have you racing cars one or more levels above you. Because the AI wouldn't stand a chance if it had to drive the same car with the same rules as you.
I for one would welcome better AI. If it means the game is going to win more. Then well, I just need to improve as well. That is what I think they call a challenge.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
PORNseek, hardware porn accelerator. Identifies and assists with porn viewing using hardware accelerated video streaming, automatic identification and filtering of media by user preference... etc, etc...
I am very sucseptible to "let's have another drink"
I for one welcome our new Artificial Overlords.. Either way, i'm reasonably sure there would be have to be some way to make the AI work in the event you don't have a card (or your slot is already occupied by a PhysX card), but given the apparently "power" of this card, i think it will be difficult. Performing things like real-time path finding during a high processor game just isn't plausable. And if they DO manage it, "Why should i pay $200 for this card when it'll work without it"
From their whitepaper:
[...]
Movement.
The Intia processor features pathfinding functionality that is both optimal
and extremely fast. Unlike today's software-based approaches (e.g.,
A*), the Intia's
pathfinding uses no heuristics, thereby guaranteeing that the optimal
path will always be found. This optimality also means that the Intia
processor avoids the common pitfalls of A*, including failures to find
a path when one exists, and the generation of "artifacts" (e.g.,
weird, unrealistic paths). If a path exists, the Intia processor will
always find it.
[...]
These guys are claiming that they can solve in a "reasonable" time any pathfinding problem they come across, without even using heuristics. If we take "reasonable time" to be bounded by a polynomial on the "grid size", then this claims means that they can solve NP-hard problems (like finding optimal paths in a weighted graph) in polynomial time.
So according to them P=NP. And we haven't even heard on Slashdot?
As a gamer would say "OMFG tey pwnd NP".
isn't line of sight calculation just raytracing?
is this the next generation raytracing card? if so, how does it compare to the CELL processor?
If hardware developers want to get people using specialist expansion cards like this AI cockup, PhysX or whatever, there are only two ways I can see to make them marketable:
- Make cards which perform in parallel for online multiplayers (MMOGs or more conventional ones), as suggested by another poster. If you can design the cards so that they benefit all players, but benefit the owner of the cards more, you have something that will gain a foothold in market perception and thus become attractive. Not sure how this would work - maybe a combined AI/PPU that improves the AI for all players, but improves the physics for the card's owner. I dunno, it's a tricky one to sell.
- Make cards which do provide some significant improvement to performance for the owner (without chewing up bus bandwidth) and stick them in the box with a couple of new genre-busting games (admittedly difficult to predict) at cost price. If people really want the game, I reckon they'll pay another £20 if they're guaranteed a performance-improving card to boot. It may be a slight loss-leader for the hardware manufacturer but, once the hardware is out there in numbers, game developers have significant reason to include support in future titles. Then you can start putting out higher-spec cards at retail price for those who need that performance edge, and hey-presto: you've created a burgeoning hardware market.
I appreciate that such cards, provided they have any market foothold, could potentially have significant impact if they're well-designed. What Aegis did, though, was underestimate the savvy of the average performance gamer. Most such consumers are tech-savvy to some degree, and are usually conscious of the relative benefits of various hardware upgrades as compared to the investment required. Aegis were arrogant and assumed we'd believe the hype, ignoring the fact that reviews, information and communities were going to be out there, and were going to be defamatory.Aegis' marketing (and cards, of course) sucked ass, and if anyone wants to follow their ideas, they'd better learn from their mistakes.
All that being said, I'm strongly in favour of dropping the whole notion and going with multi-threaded games development to match our ever-cheaper multi-core CPUs. All I'm saying is, if these AI guys want to succeed, they're going to have to be really fucking canny and stop treating the games market like a pool of irresponsible cash. Not many of us live in the basements of our parents' mansions...
Meta will eat itself
ATI was bought by AMD. That was part of the reasoning for the all in one CPU. Imagine the price drop if you could just add a GPU core to an AMD 64 X2G chip. Heck, video RAM could just be another slot on the motherboard. Want to upgrade your video to 512MB? No problem, just buy a DIMM...
In truth, I must confess this isn't even that original a thought. There was speculation about AMD's multi-core approach as soon as they announced the architecture in that you could have a couple of general purpose CPU type cores, coupled with special DSP, GPU, or other cores, all working at HTT speeds.
The cesspool just got a check and balance.
"Is this the 'way of the future' for PC titles? Will games powered by specific pieces of hardware become the norm?"
More questions:
Will game developers be willing to optimise their games for people own specific gaming computers containing this hardware only and ignore "average" gamers with "average" computers? Or will games be made for both scenarios, and if so, how different would gameplay be on different hardware? Will exclusively optimised games be very expensive to justify an ROI from a smaller market?
RTFM; please, I beg you.
Most games are about what you can do, and what you cant do.
Theres a Quake1 engine that able 10.000 monsters in a room. But is unplayble because is CPU limited. If you can able 10.000 monsters, you can have Quake Wars:Enemy Territory with a real-size army of enemys. 8000 instead of 80.
And the game will FEEL bigger and deeper.
This enhance in a way that enhance even NEW gameplay styles. Check Spellforce if you can.
-Woof woof woof!
> Having worked in the industry on a MMORPG, I agree that it would be nice to use client machines for extra distributed processing, but there are issues.
:-)
>
>First, as a rule of thumb in multiplayer development, never trust the client machines for anything other than controller and view data for that player.
indeed
[..]
>Second, although bandwidth is going up, having a remote machine control AIs in-game introduces two issues (off >the top of my head). First, if the client machine hosting the AI goes down or loses its connection to the server, >the AI will become "brain dead" to all other players in that general area, at least until a new AI controller can >be spawned somewhere.
You can fix that with 2/3 users controlling the exact areas. So if one die, you use the other.
If you have more than 2 users controlling a area, you can validate his output.
> As well, that AI has then lost its memory of everything it has experienced up to that point. Second, although >bandwidth has increased, latency is still your issue. No longer are state updates for your NPC coming only from >the server, they are coming from some remote machine, hitting the server, being validated (if the server is smart at all) and then being propagated to all other players in that area. That introduces a validation check and a >second hop, which will slow down AI responsiveness.
You can have 2 AI: the strategist, that can run on your grid of clientside machines. And the tacticalist, that run on the server. The strategist manage strategic stuff (I need food) and the tactical inmediate stuff.
You dont need the same timesteps.
> We thought about this a fair bit where I worked. We decided that it just wasn't doable.
> Nice dream, though.
I dont like the idea because will be complex, but is posible. Complex mean easy to broke.
-Woof woof woof!
Why not produce a single fast FPGA addon card and reprogram it for whatever means we want?
I need physics accelerator? -> I load the physic accelerator program.
I need AI accelerator? -> I load the AI accelerator program.
I need sound acceleration? -> ditto.
I need custom accelerator? -> I program and load my own bloody one.
You can even switch them at run-time, if you want to.
FPGAs these days aren't expensive and are FAST. We can easily get hundreds of thousands logic cells running at over 500MHz.
The dumb AI is mandatory part of most games.
For most of the games out there, people don't want good AI. Good as in the AI makes a decent job of finding the best way to defeat you.
If the ghosts were really smart in Pacman, it would be really really hard to get very far. Imagine the ghosts just hogging the last pill once you've run out of power pills.
If the enemies were really smart in DOOM and other FPS, you'd probably never make it past the first few levels.
Imagine if mobs behaved really intelligently in WoW or some similar game.
In many games it is fairly easy to come up with something (even with minimal learning - just decent heuristics) that beats most if not all human players.
Most humans are stupid/ignorant, don't have perfect aim, reflexes and timing so for most games out there, they wouldn't want to play against a smart AI.
Even in chess you have settings for weak human players.
Otherwise most human players would be crushed, and the game just wouldn't be fun for most.
The only fun part then could be writing/designing bots to battle each other and possibly watching them in action.
From their site.
'Unlike today's software-based approaches (e.g., A*), the Intia's pathfinding uses no heuristics, thereby guaranteeing that the optimal path will always be found.'
That doesnt make any sense. A* guarantees an optimal path _because_ it uses heuristics. A* may not be the greatest algorithm computationally but it will always find the best path. (Its admissable)
'This optimality also means that the Intia processor avoids the common pitfalls of A*, including failures to find a path when one exists,'
That isnt right either. If there is a path and it is the best path thats the path A* will return.
Whats more, A* is horribly out dated and should be fairly well modified to make it real time. (RTA*) Now I have written, and I wouldnt be even remotely surprised if this was a case for some commercial games, a real time A* program that simply re-tests the A* path each turn. (Horribly, horribly, inefficient.) This is what the article implies is going on and it just shouldnt be happening. We should have progressed enough with AI to implement a proper real time pathfinding algorithm.
'Most importantly, this very significant speed increase gives the Intia processor the adaptability to support large, dynamically changing maps.'
Once again a real time algorithm should be able to deal with any dynamic map behaviour and the size of map shouldnt be an issue with some decent discretizing. (Discretizing, which will be essential otherwise youd flood the memory no matter how fast your AI can be processed especially without heuristics.)
'such as the need to find a path that passes through certain locations (e.g., hiding points, enemy positions). '
Just apply the right costs to the nodes in an A* algorithm (Which you can do in real time.), itll do the same thing.
'Sensory information'
Now we're getting somewhere. Line of sight tests on that quantity of agents in such a short time is fairly good going and quite helpful.
There are once again software tricks to get around this fairly effectively by simply passing information to the agents that they shouldnt really have, but they can lead to problems such as enemies spotting you through several layers of brick wall.
'Terrain Analysis'
As pointed out previously dynamic terrain analysis can be done using RTA* algorithms with very little problem. It shouldnt have a major impact on the processing because its not changing the AI its changing the world. (Which has to be updated constantly anyway.)
The reason NPC's get stuck on walls in games isnt because of A* its because of bad AI. From what they say on there site they arnt really going for making AI any better. Just doing what we already should and could be doing already.
Their insistence on not using heuristics is mystifying. Nearly every AI problem in existence needs good heuristics else it will chew through gigabytes of memory and hours of processor time to complete its task.
Just as an example. The number of moves in chess without heuristics in the first turn is 2048. With heuristics its 40.
Not to mention that throughout the article they seem to link heuristics to inadmisable algorithms which is nonsense.
It has been a while since ive done AI so someone may well prove me wrong, but from what I can see this chip just seems to use the same old algorithms to produce good AI through brute force instead of intelligent design.
This isnt like physics. If you simulate one brick falling and want to simulate 2 you double the power, simulate 3 triple it, easy. In the field of AI. That isnt going to cut it, the problems scale up in memory and processor requirements exponentially this kind of thing is only delaying the inevatable. AI _needs_ more research not more power.
I tried to sell this idea in 2000. Everyone I talked to (Id, Apogee, Blizzard...) felt that the processor was capable of doing the work, an API would restrict developer decisions on which algorithms to use, and the value/cost wouldn't be within comsumer ranges. Not being one to patent every idea just in case someone else makes a success out of it, I wish them luck. I do, however, hope they open source the drivers. I started some GUI interface AI code for Enlightenment and would like to dig it up and work on it again.
Having to work for a living is the root of all evil.
Trying to sell solutions like a "Physics chip" or an "AI chip" is the joke. They are trying to sell you what could be a very generic add-on product but they are shoe-horning into doing a very limited set of things. They do this so they can charge developers an arm and a leg for the API, and so that they can make you buy three or four accelerator cards to use up all those empty PCI-e slots (thanks to motherboard integration of everything).
No, the answer is simple multithreading and taking advantage of dual core machines (or quad core when that day comes around).
I could maybe see the use of a parallel/vector processing add-in card maybe coupled with an FPGA that developers could use to tackle specific embarrasingly parallel problems for the game code. The card would be multi-purpose and it'd create a new market for developers to come up with FPGA code that does whatever hot new thing might need accelerating:
1) Water and cloth simulations
2) Procedural texture generation
3) Radiosity calculations
4) Swarming/agent-based modeling
You don't need seperate cards to do this stuff. One card is fine. Two for extra oomph. Or in the case of AMD w/hypertransport, you could have special purpose plug-in coprocessors.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
Bet you 10 bucks that this AI accelerator is a PowerPC mated with a Xilinx FPGA.
Same thing for the PhysX processor.
Hmmm... i wonder if you could make one into the other with a firmware flash and FPGA update.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
Maybe now Rainbow Six will actually be playable because one's teammates won't always do stupid stuff and die (or get stuck in a corner).
Error 404 - Sig Not Found
Is this the 'way of the future' for PC titles? Will games powered by specific pieces of hardware become the norm?
Amiga