• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Why the PPU is a fantastic idea

C

Chocolate Pi

Senior member
Some people just don't like change. The hypocrites will take pride in their deticated GPU, but are strangely hostile against the idea of a deticated Physics chip. "Why do we need another thing to buy?" they ask. "What's next, a deticated AI processor?"

One needs to only think clearly for one second to see how silly these arguments are. Why do we need GPUs? After all, with multi-core processors, surely one of the cores could handle graphics, right? Of course, we all know that GPUs are masters of what they do. We have industry standards for graphics as we all know. Graphics cards are optimized to be incredibly efficent at DirectX and OpenGL, as well as having pipelines designed exclusively for the best texture performance.

These standards are beginning to emerge in the physics programming world as well. More and more games will be taking advantage of Novadex and Havok in coming years. We have finally reached a time where we can have deticated hardware to accelerate the performace based on these standards. Deticated, focused physics hardware can do hundreds of times what any modern CPU can do.

Think this is "marketing hype"? Go watch the demos and listen to the developers.

Ah, but you claim that CPUs will soon get better and the multi-core CPUs of coming years will eliminate the need for deticated hardware? LOL, technology gets better over time, who would have thought!? Well, I've got news for you buddy. The PPU will get better too. As the CPU goes multi-core, it will too, just like the GPU will. Deticated hardware will ALWAYS be two steps ahead of general processing.

And for some reason, someone likes to bring up that the Xbox 360 and PS3 are going in the direction of everything being pushed onto the CPU. Well, DUH. They are ~$300 consoles, bound by budget. Their design goal is consolidation; add-in parts are not good for the mega-mass-production console market. The PC market is not bound by these constraits, and hence always stays ahead of the console market despite the latter's optimized code.

In conclusion, the PPU is a great idea. The challenges it brings to the industry are the same as those of the GPU, but the incredible performance benefits are similar as well. And maybe, one day, we will get standardized AI middleware platforms, and we will indeed need deticated, optimized AIPUs.

 
I didn't see this thread earlier, but I agree with you 100%. 🙂

I think the people who say that multi-core CPU's make PPU's irrelevant don't realize how much their system would be brought to a crawl if the CPU tried to do the kind of stuff on the scale that the PPU is proposing to do. 😛
 
its not a matter of scale that matters. if the multicore cpu can handle even a bit of what the ppu can, its probably good enough for most. it comes free.. or will soon enough when dual core is standard. its a matter of not being willing to pay for yet another expensive addon. sound and video clearly have their uses in normal use. ppu is games only and harder to justify. it would have to be quite a bit cheaper
 
I think the PPU just isn't as tangible as a 3D accelerator, at least not for now. The transition from 2D to 3D accelerators was easier because video cards existed before 3D accelerators were around in the consumer market space. Also, the existed in the workstation/professional market space, with a well known API(OpenGL). PPUs are currently nowhere to be found, since everything to do with physics is currently done in software. It's essentially trying to go from zero to hero. Also, the benefits of better physics at this point are clearly limited to game applications, specifically FPS games. It doesn't benefit anything else AFAIK, and even if it does, the benefits are small. As the architecture is better known, the PPU may be able to perform other tasks, who knows? But at the expected time of its introduction, its use is limited. I hope it succeeds though.
 
i dont know about anyone else but i hope this ppu stuff takes off. asus afaik is onboard already with it. my only concern for now is the price.
 
Improving a CPU improves all-round system performance in general.
Increasing RAM improves performance in most cases.
Utilising a more powerful GPU improves gaming performance.

What does a PPU do for games?

No-one can answer this fully yet since I dont believe there are any games or programs at the average consumer level that take advantage of a PPU.

I can speculate about how they might be used and I have done in other PPU threads. I think the main problem people see (or fail to see) with a PPU is what it is going to do to games. Would you really notice realistic grenade bounces whilst playing UT2k4? or notice the rockets in Quake 4 miss you because the wind shifted their direction slightly?

What Im really trying to say is CPU/RAM/GPU provide people with improvements they can feel and in a GPU's case, may see as well. At the moment, the improvements made by a PPU arent quite visible or tangible yet.
 
With a PPU, in the future we'll probably see better physics operations in games and will offload quite a fair bit of the work off the CPU, leaving it to manage the AIs, etc. etc.
 
I think they'll be a flop for a number of reasons:

1) There's no game support yet.

2) I've never thought to myself, "I wish those physics looked better." In fact, I'm more than satisfied with the physics my CPU processes for me right now in games like HL2, for example.

3) It's too expensive for what it currently does (not much due to game support). Price it as cheaply a sound card (~$40-$70), and sure, I'll try one out and see what I think. I'm sure as heck not going to spend $300 on something that I can hardly use.
 
I think it will be a flop as well.

Simply because of the trend of computer parts to become more consolidated. If anything happens with this PPU thing, it will be a chip integrated into the graphics card.

Also the market for a PPU is even smaller than for GPU's. I don't think there will be much money in it.

However I disagree with Gerbil333, the physics of HL2 really made it a great game. While playing Doom3 I did ask myself, are there any physics in this game?
 
One day I decided to play CS:S. I decided to spawn 20 bots with only knives on the opposing team. I climbed on a roof and watched the silly bots gather in a big mob underneath me. Then I decided to start executing them with headshots. After killing 10 bots their silly ragdoll bodies piled on top of each other, and the fps dropped from 60 to 1.
The moral of the story: software-based physics are quite pathetic compared to hardware-accelerated physics, and after 5 years when playing HL3 and piling up combine bodies to build a stairway to heaven, we will all collectively laugh at the silly people who thought dual-core would be enough for physics.
 
The price definately needs to go down. I'm not even going to buy a $300+ video card, so there's no way I'd spend that much for a PPU card.
 
I don't see the PPU as a flop, but not something to see soon in PCs. Before it gets to PCs, I have a feeling Intel and AMD will get real vector units on their x86 cores. The Cell and PPU have their place, just not for PCs or your typical server. If the Cell could get a real G5-comparable PPC part on it, or the PPU design were licensed to Intel/AMD/IBM, there will be a place for them for normal users.
 
PPU would br a great idea, but I agree with the others who stated that since it's made for games ONLY, it would best belong ATTACHED to an appropriate mid-to-high-end 3D video card. Perhaps even two levels of PPU, low-grade for mid-range cards, high-grade for the video cards that currently cost more than a month's rent/mortgage. 😉

Low-end cards? No PPU for you! 😛
 
Wait and see approach for me!
(mad rhyme!)
People definately have some lofty expectations for this PPU. Unless games can be designed to run decently with or without a PPU, and cheaply at that, I doubt Ageia will enjoy much success.
But hey, people are willing to buy $500+ graphics cards so what's another $300?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top