Physics and AI chips

[Gogf]

http://techreport.com/onearticle.x/8360
http://www.extremetech.com/article2/0,1697,1855137,00.asp
http://www.pcper.com/article.php?aid=140&type=expert

Imagine, the year is 2015. Your computer comes with an AI and a physics chip standard. This allows you to not only play complex games that involve CPU intensive calculations for a competitive AI, and a realistic physical enviroment.

Well, maybe the above could have been phrased a little bit better, but you get the idea . What do you think about this? Is this a good path of development for the future, or a useless, if somewhat interesting, dead end? Will these chips revolutionize gaming in the future?

 

[GenMarshall]

Ok, what would this computer do to my Microsoft Train Simulator?

 

[Speedo]

It has great potential, but the problems are (1) convincing developers to invest time and $$$ into programming for the PPUs, and (2) convincing consumers that there will be games which take advantage of the PPU, and the advantage over a non-PPU accelerated game will be worth the price of the card.

I think it will be a while before it really takes hold, if it does.

 

[Chieftess]

Ok, what would this computer do to my Microsoft Train Simulator?

- Control fuel usage based on the effects of gravity. If you're going downhill, you can go faster but not use up as much fuel as opposed to uphill.

- Manipulate terrain based on natural changes in geography, climate, etc.. (i.e., too much snow will trigger an avalanche. Too much rain could trigger mudslides, wiping out the track).

- Have more realistic fuel-to-energy costs.

- More realistic train speed based on a variety of factors in the physics engine.

Those kinds of things.

It has great potential, but the problems are (1) convincing developers to invest time and $$$ into programming for the PPUs, and (2) convincing consumers that there will be games which take advantage of the PPU, and the advantage over a non-PPU accelerated game will be worth the price of the card.

I think it will be a while before it really takes hold, if it does.

I can think of one thing - N64. It had superior graphics (if you can look past Super Mario's pink sky -- by potential, I mean things seemed smoother than the Playstation) potential compared to the Playstation, but there were two problems:

1 - Space was limited to one cartridge.
2 - It was hard to program for.

The 2nd will be interesting to see how it goes with the chips.

 

[Speedo]

- Control fuel usage based on the effects of gravity. If you're going downhill, you can go faster but not use up as much fuel as opposed to uphill.

- Manipulate terrain based on natural changes in geography, climate, etc.. (i.e., too much snow will trigger an avalanche. Too much rain could trigger mudslides, wiping out the track).

- Have more realistic fuel-to-energy costs.

- More realistic train speed based on a variety of factors in the physics engine.

Those kinds of things.

If you can convince M$ to go back and invest the man-hours to add those things, that is

No doubt the potential is mindblowing... some of Aegia's demo videos are pretty amazing. But convincing game developers to truly put it to use...? Well, we'll see.

The 2nd will be interesting to see how it goes with the chips.

I don't thing it's the difficulty... after all, anyone's who's had a calculus class should be able to do most of the math. It's more the complexity. I mean, imagine something as simple as a simulation of our solar system (major bodies: sun, planets, moons). Between calculating the gravitational forces, updating momentum, calculating velocity from that, and updating position, you're probably talking about 15,000-20,000 calculations for each of your deltat "steps." And that's for a mere 130 or so objects.

But that aside, the other thing I've read is that, at least initially, the Aegia PPU will only support a physics engine supplied by Aegia. I think that will probably change with time though.