AGEIA is a bust.
Ever since AGEIA announced the PhysX Physics Processing Unit some years ago people (including our CEO) have bandied the idea about. Everyone wants this to seem cool right, I mean c'mon more hardware, dedicated to physics even, MUST be a great thing. Some of the braver ones have even bothered to look me in the eye or write a dev question and ask abot the subject. My general reply is to look at them as if they are completely out of their mind, give them a doctor spock-brow and slowly utter the most confusing word they appear to have ever heard. "No."
Occassionally we then go into the details of why a seperate physics card is moronic. Market penetrationg, inability to run actual game play physics on it in a multi-player environment, and the little fact that 2 giant companies already make add in cards for your PC and they are some of the most insane hardware out there and they have more spare processing power than HAL just lying around for things like physics if it was a good idea or when it becomes a good idea and besides, seriously, do you think for a second that Nvidia is gonna let some n00b roll into the playground and muscle in some $200 add on cards. Yah! Uki. kkbye.
Link
Occassionally we then go into the details of why a seperate physics card is moronic. Market penetrationg, inability to run actual game play physics on it in a multi-player environment, and the little fact that 2 giant companies already make add in cards for your PC and they are some of the most insane hardware out there and they have more spare processing power than HAL just lying around for things like physics if it was a good idea or when it becomes a good idea and besides, seriously, do you think for a second that Nvidia is gonna let some n00b roll into the playground and muscle in some $200 add on cards. Yah! Uki. kkbye.
Link
9 Comments:
Well, not just Nvidia but also Havok.
I think it would have been a great idea 10 years ago but not today. There's just too much room to play with in the CPUs of most games today, especially with the high probability that multi-core processors will likely become the norm over the next 5 years.
I have seen Ageia at work in XSI and it is quite impressive. One of the things that excites me a bit is the prospect of running a simulation in XSI and outputting that to a keyframe animation. I know that I would love to use it to animate particles at the very least.
I think Havok has the right idea. We will most likely see Havok be an even bigger player in the next year while Ageia finds itself relegated to a simulation environment. The only way that would change would be for Ageia to get the price for their card down to <$50. I have no qualms believing that a big title game could be packaged with an Ageia card for $100 and that people would buy it in droves. It would have to be a huge title though.
I had a 4 meg Monster 3D card back in the day, as a second to my 4 meg Diamond Viper.......
Guess it was an early physics card.
Merc88
How do much can I get paid from Nvidia to slander Ageia. With an unbiased view, Ageia card is really impressive.
Yeah, a physics card wouldn't work in a player vs player game. All you can do is add pretty effects for people that have them. You can't change the game in any way. You can't, for instance, have people with physics cards generate 10,000 pieces of shapnel when their grenades are thrown vs 1,500 for players who don't have physics cards.
I don't think the point was to slander them but to point out that the guys from Havok don't seem all that impressed and also to note that as a business model there are 2 add-on card companies that pretty much have the market sown up who could very easily add physics chips to their existing graphics cardws.
Only time will tell if the Ageis card gets any traction but it is a tough road to hoe. Making things pretty is a very solid goal for any game but more games are going multi-player and in multi-player you have to limit the physics co-processor to just making more particle effects and that might not be enough to ensure the market penetration needed. To boot, CPU's are going dual core and graphics cards are insanely fast these days. The need just might not be there.
I think the 3d card wars are a perfect example. Nobody reallly knew they needed a 3d card for a long time and then everyone needed one yesterday. Physics co-processors may well be the same but 3dfx is now only a fond memory.
3D graphics, 3D sound, 3D physics... Each having to have a copy of the 3D world model in their own custom, finely tuned format.
That's a lot of pushing the 3D world around.
And meanwhile it's not as though the CPU doesn't need its own copy(s).
What would truly rock would be a 3D subsystem that can either run all of these systems in software or hand them to cards as you have them available. So that the CPU only has to push world data once, and can make 3D queries in a consistent fashion like it can to the graphics card.
Maybe creative ought to get with nVidia and come up with a spec for bridging the sound and video cards.
Why would Havok, as a direct competitor, praise AGEIA? It's in their best interest to give them as much bad PR as possible.
Setting aside the problems physics cards might bring for decentralised multi-player games, it should be pretty clear that improved physics is something that is very desirable from an immersion point of view.
Havok actually proved exactly that when they entered the market.
Purpose-built hardware is always a bit silly and I doubt it would reach mass market-penetration at >$200 prices. But once there are a couple of games that make use of it, and the prize if not completely ridiculous, you can be sure that the 'insane gamer segment' will be buying those cards, just like they buy any other expensive, hyped-up product that might make their gaming experience just a little bit more shiny.
Once a certain baseline is reached I would actually expect this to be something that moves onto mainboards (or possibly onto graphics cards) and not be sold as a separate card. But since the appeal is pretty big I don't think it will go away.
I'm not quite sure why you say that existing GPUs have much processing power to spare. How so?
Why would Havok, as a direct competitor, praise AGEIA? It's in their best interest to give them as much bad PR as possible.
Havok isn't exactly a competitor as they don't make any hardware.
I'm not quite sure why you say that existing GPUs have much processing power to spare. How so?
The general consensus is that a single moder gfx card such as the ATi 1900 XT has more than enough horespower for any game on the maerket at up to normal 19" flat panel native resolutions that within a couple of years there will be so much latent horsepower on a modern graphics card that it could be used for physics or sound. Take into account dual or quad processors and another card that has to be upgraded just isn't going to make it.
Havok isn't exactly a competitor as they don't make any hardware.Havok isn't exactly a competitor as they don't make any hardware.
They provide a physics API, Ageia provides a physics API. The functionality they aim to provide is targeting exactly the same goals.
As I said, I doubt we'll see much stand-alone PPU cards. They will go onto mainboards or graphic cards at some $10 in added production cost. Or bust.
I've not had one graphics card that went underused. I fail to see why this should change significantly.
But even allowing that machines become so powerful that a cheap, purpose-built chip would not be useful, Ageia is currently developing for those resources to be available whereas Havok is working with the hardware available.
I don't know where this is going but here's some guesses:
* Ageia's main income will be from patents
* Ageia may make some money from mass-producing chips for mainboards or something but I don't know if there is actually money in these areas
* A generic physics API will become part of DirectX
* Both Havok and Ageia provide some SDKs/Engines, like what is happening in the 3D world.
Well, the latter point is already happening, with Ageia being significantly cheaper, I believe?
I don't really care either way, as long as I can blow stuff up more convincingly, at some point.
Post a Comment
Links to this post:
Create a Link
<< Home