Gamespot Interview with AGEIA

[onetimeposter]

Q&A: Ageia CEO Manju Hegde
CEO talks about the PhysX processing unit, the PlayStation 3, and getting developers to support the Ageia PhysX PPU.

Physics technology startup Ageia has had a busy summer getting its PhysX software development kit into the hands of anyone who has come within five yards of a game-development workstation. Mythic Entertainment, Quantic Dream, and Digital Jesters have signed on in recent weeks, joining the growing ranks of PhysX SDK developers, which already include Epic Games, Ubisoft, Cryptic Studios, and Ritual Entertainment. Earlier this year Sony announced that it was including an optimized version of the Ageia PhysX SDK with the PlayStation 3 SDK.

Developers can use the Ageia SDK to add physics effects and interactive gameplay elements to their games. While one imagines that most developers will be using the SDK primarily for CPU software acceleration without a dedicated physics processor, games using the Ageia PhysX SDK will have built-in support for physics hardware acceleration. Ageia must have this application support to give gamers a reason to buy a PhysX PPU card.

GameSpot caught up with Manju Hegde, Ageia founder and CEO, for an e-mail interview to discuss the current status of the PhysX processor and his predictions for physics-accelerated games.

GameSpot: Sony announced last month that it has sublicensed the Ageia PhysX SDK for the PlayStation 3. How does getting your physics library into the Sony PlayStation 3 SDK help Ageia on the PPU front?

Manju Hegde: The Sony announcement is significant for two major reasons: More developers will be exposed to the powerful features, particularly multithreading, within the Ageia PhysX SDK; games that are developed for the console will be easily ported to the PC platform thereby creating a larger installed based of Ageia PhysX hardware-accelerated games. Finally, games on the PS3 are likely to incorporate physics in gameplay, which makes physics an essential and vital component of the game, thereby paving the way for PC games which the physics processing unit is essential.

http://www.gamespot.com/news/2005/09/01/news_6132402.html

BOLD Text: the CEO of the company doesnt know how well the SDK will work will cell, i wonder how they know it wont work with Xbox 360 (Extremetech)


[Forum Rule : DO NOT COPY/PASTE WHOLE ARTICLES !!!]

 

[Nicked]

Wasn't your other account banned? Now you are back looking to start arguments.

 

[onetimeposter]

Wasn't your other account banned? Now you are back looking to start arguments.

What are you on about?

and besides. im trying to correct misinformation not spread it. prove me otherwise

 

[Nicked]

Funny, I got the distinct impression that your only purpose was to spread misinformation and propaganda with your old account.
Its quite obvious when you make a thread yet don't discuss the content you yourself post, only want to instigate arguments as to whether or not Extremetech was lying (hint; they weren't!).
Oh, and weren't you warned about posting complete articles? I think you were.

The article itself is interesting, especially the launch details. And I couldn't agree more with this:

Next-generation games need to go beyond simple effects of boxes, and that's what PhysX is all about.

I doubt most next-generation games will rely on physics due to reliability/unexpected oucome issues, but I have high hope that there will be a few stellar games that really prove that physics can be used to greatly enhance gameplay.

 

[mckmas8808]

I doubt most next-generation games will rely on physics due to reliability/unexpected oucome issues, but I have high hope that there will be a few stellar games that really prove that physics can be used to greatly enhance gameplay.

Why are people saying this? Are you people looking for the worst just to receive something better? You should know by know that Sony will be pushing what physics is all about. They showed demos at E3 to prove it.

 

[Qroach]

Why are people saying this? Are you people looking for the worst just to receive something better? You should know by know that Sony will be pushing what physics is all about. They showed demos at E3 to prove it.

God, would you pipe down... "Sony will be pushing what physics is all about" some of the statements you make are very, ummm, let's just say you don't know how physics will be used in games aside from rag dolls and boxes/spheres moving around.

 

[ERP]

Why are people saying this? Are you people looking for the worst just to receive something better? You should know by know that Sony will be pushing what physics is all about. They showed demos at E3 to prove it.

The biggest issue with physics in games, is that it's unpredictable.
I hit the boulder 1 inch to the right and the avalanche I cause though spectacular doesn't clear the entrance to the cave the designer intended and I have to restart.

In the short term at least it'll be more like lens flare, the leatest cool thing with little impact on gameplay.

A number of people have tried to build games around physics over the last 5 or so years, none have really worked. It's somewhat odd really dropping a few "physics" objects in a world and playing with them is fun all on it's own, but no-one has really managed to extend that fun into a game without it simply becoming frustrating (tresspasser anyone?).

 

[Jawed]

Firing tables at zombies is awfully good fun.

An explosion that just coincidentally blows a really nice gun in your direction so that it lands at your feet is cool.

etc.

There's plenty of fun to be had with physics effects in games - but the type of game is what really determines whether any kind of physics effects can play a part in gameplay as opposed to merely being eye-candy.
Jawed

 

[Shifty Geezer]

I can't imagine it'd be that hard to design aspects incorporating physics as long as provide 'more than necessary' to solve the task. eg. Instead of using an avalanche use some explosives by the entrance. Or have a scripted event that if an avalanche occurs near enough the entrance clears as well. I would have thought the main reason physics based games haven't taken off is the sheer processing demand to get anything complex going with more than a few ragdolls or couple o' dozen piled boxes involved.

What I'm more concerned with is how physics is going to be accomodated in MMOG's? How does syncing work on massively complex simulations? Even two-player poses a conundrum with complex physics. Consider the E3 bathroom demo and those ships sailing around. Imagine that's a 2 player game over the network, controlling those jaunty little vessels and trying to sink each other, while waves are applied to the bath tub to make a rough sea. How readily can the two water models on the two different PS3's be synchronized so both machines have the same waves in the same positions? You'd be looking at lots of network traffic it seems to me and all the associated problems that has.

 

[Phil]

I'm not quite sure how you'd approach a problem like that, but I'd assume you just send the player's interaction over the network and let each PS3 simulate the behaviour individually? Given the same "cause" - the "effect" will be the same on both machines. The only thing that would cause problems is that physics run at a much higher framerate and the dataflow between the two networked consoles would have to be fast enough for both to interact in "real time"...

 

[seismologist]

Going back to this motorstorm video
http://streamingmovies.ign.com/ps3/article/640/640147/motorstorm_080805_wmvlowwide.wmv

I think the mud effect is a perfect example of physics (particle fluid dynamics) in action.

 

[Shifty Geezer]

I'm not quite sure how you'd approach a problem like that, but I'd assume you just send the player's interaction over the network and let each PS3 simulate the behaviour individually? Given the same "cause" - the "effect" will be the same on both machines.

I don't believe the maths invovled is that accurate to produce the same results from the same beginnings. I as there's rounding errors and a simulation engine may well use inaccurate shortcuts to speed things up, consistency is lacking in most physics engines from what I know.

 

[PG2G]

I don't believe the maths invovled is that accurate to produce the same results from the same beginnings. I as there's rounding errors and a simulation engine may well use inaccurate shortcuts to speed things up, consistency is lacking in most physics engines from what I know.

Well, wouldn't they be the same rounding errors and the same inaccurate shortcuts? Given the same input, you'd assume the engine would produce the same output.

 

[ERP]

I can't imagine it'd be that hard to design aspects incorporating physics as long as provide 'more than necessary' to solve the task. eg. Instead of using an avalanche use some explosives by the entrance. Or have a scripted event that if an avalanche occurs near enough the entrance clears as well. I would have thought the main reason physics based games haven't taken off is the sheer processing demand to get anything complex going with more than a few ragdolls or couple o' dozen piled boxes involved.

What I'm more concerned with is how physics is going to be accomodated in MMOG's? How does syncing work on massively complex simulations? Even two-player poses a conundrum with complex physics. Consider the E3 bathroom demo and those ships sailing around. Imagine that's a 2 player game over the network, controlling those jaunty little vessels and trying to sink each other, while waves are applied to the bath tub to make a rough sea. How readily can the two water models on the two different PS3's be synchronized so both machines have the same waves in the same positions? You'd be looking at lots of network traffic it seems to me and all the associated problems that has.

I saw demos 5 years ago of characters that stood and walked using physics (they were primitive but they worked). I saw fluid dynamics and turbulence modelled on Pentiums in the 1GHz range. Playing with the demos and applying forces to things was a lot of fun.

The math power has been there unless you want to stack things up excessively. The games haven't, physics is physics, it's inherently unpredictable in that small changes in the input forces can have massive effects in the final outcome. That makes it difficult to test and verify.

As pointed out throwing a table at a zombie, can be cool, and you can limit it to add to the coolness factor without breaking the game. But you can't really use it ubiquitously, when you commit something to be driven by physics, you have to design around the consequences.

Doing it in a MMORPG isn't necessarilly impossible, but it is potentially extremly difficult to get right. Ignoring PC's and the wonders of DX randomly changing your rounding mode, a good physics simulation will be consistent run to run for a given set of inputs. But you're still stuck with dealing with whatever the lag has introduced on the various clients, but you have to deal with that for everything else anyway.

 

[ERP]

Well, wouldn't they be the same rounding errors and the same inaccurate shortcuts? Given the same input, you'd assume the engine would produce the same output.

Assuming no side effecta and no internal state. This is true.

However last time I looked on PC's in particular it isn't, DX plays with the rounding mode on the processor, so in effect your graphics can change the outcome of your physics simulation.

 

[Titanio]

Assuming no side effecta and no internal state. This is true.

However last time I looked on PC's in particular it isn't, DX plays with the rounding mode on the processor, so in effect your graphics can change the outcome of your physics simulation.

Is this as much of a problem in the console space?

It reminds me of Kutaragi's comments about sharing results between RSX and Cell and having "synched" rounding modes between the two etc.

 

[Shifty Geezer]

Assuming no side effecta and no internal state. This is true.

However last time I looked on PC's in particular it isn't, DX plays with the rounding mode on the processor, so in effect your graphics can change the outcome of your physics simulation.

Okay. It didn't make much sense to me why given the same conditions and same processing algorithms you could end up with different results, but I know from my own experiences it can happen. Didn't realise it was DX changing things. That at least suggests sync'd simulations between machines are possible.

 

[aldo]

Firing tables at zombies is awfully good fun.

An explosion that just coincidentally blows a really nice gun in your direction so that it lands at your feet is cool.

etc.

There's plenty of fun to be had with physics effects in games - but the type of game is what really determines whether any kind of physics effects can play a part in gameplay as opposed to merely being eye-candy.
Jawed

http://www.ageia.com/press/physx_ca...sh_lavajets.exe
http://www.ageia.com/press/physx_ca..._napalmlong.exe

Imagine how cool it would be to see these effects used for a weapon such as Unreal Tournament's Biorifle. Splashing shots off ceilings and walls to wash enemies with bio material or napalm in hidden areas and suffering the consequences if you fire with too much pressure and hit surfaces deflecting material back at you. Would also be a blast directing a continous stream onto an enemy squad or on dense vegetation to make a clearing.

-aldo.

 

[mckmas8808]

Respectfully speaking

God, would you pipe down... "Sony will be pushing what physics is all about" some of the statements you make are very, ummm, let's just say you don't know how physics will be used in games aside from rag dolls and boxes/spheres moving around

Oh please stop trying to down everybody that cares about physics in games. I'm going to quote something that I posted in another forum.

I'm going to give you some of the clips that show how physics can change a game. Graphics wise and gameplay wise. What I'm going to allow you to see everybody here will agree with me on. And hopefully you will too.

This is the FF7 tech demo
Look at Aeris' hair as she is squatting down. The hair moves more when the wind picks up. Look at her dress as she stands up. It falls realistic physics. Look at her dress as see walks. It moves with her legs. Look at the flowers moving with every stride.

http://streamingmovies.ign.com/ps3/article/641/641244/final_fantasy_tech_demo_081105_wmvlowwide.wmv
I know you liked that.

EA's Fight Night Video
I don't even have to explain why physics plays a big role in this video. It basically speaks for itself. Just check out the boxers faces as they get punched.
http://streamingmovies.ign.com/ps3/article/640/640134/fight_night_080805_wmvlowwide.wmv
Motorstorm
Again another video that shows how how tens of thousands of small particles can change a game. Look at the end when the yellowish greenish car's front end breaks into pieces. Thats physics at work man.
http://streamingmovies.ign.com/ps3/article/640/640147/motorstorm_080805_wmvlowwide.wmv
Nice right?

Duckies Demo
Look at the ship's sails as the cannon balls fly through them. You don't think that could be used as gameplay? Make sure to watch in the when the guy uses the eyetoy to pick up water with a real life cup. Watch for the physics in this demo. It's lovely.
http://media.ps3.ign.com/articles/6.../615/615000/sonycon_demos_duckies2_wmvlow.wmv
Exploding gas station
This is a gas station exploding in the middle of the desert. Check out how much physics calculations have to be involved to do something like this. People would love to destory a building like this and have it explode in a realistic fashion. THIS IS PART OF A REAL GAME TOO!!!
http://media.ps3.ign.com/media/748/...14/614835/sonycon_demos_gasstation_wmvlow.wmv

Q just because you have such a limited scope on how physics can change how videogames look and play doesn't mean everybody else does too. If you are a dev I hope your fellow employees help you out in the future to open your eyes.

 

[Jawed]

None of those game demonstrations show how physics affects gameplay. Unless there's a chance that she'll trip over her dress or she'll be so busy brushing the hair out of her eyes...

They're all eye-candy.

As someone said earlier, the equivalent of lens-flare.

Jawed