Getting PhysX-ical

Getting PhysX-ical

By Ed Fear

July 30th 2007 at 1:25PM

Introducing a solution that brings new way of thinking about physics may be quite an achievement in itself, but Ageia isn't resting on its laurels: its next job is to convince developers that hardware-accelerated physics is the future. Ageia's VP of marketing Michael Steele tells Ed Fear how and why...

What has Ageia been doing in the past year to help get developers onside?

The issue for us is the proof being in the pudding - we need to show games with the Physics Processing Unit making the difference.

We have to work very closely with the developers - guys like NetDevil, Epic and UbiSoft. We have to tie our product, our middleware in with the right kind of tools - things like Alias and Softimage. We also have to do some things with the game engines themselves - the engines, it turns out, require a fair amount of work to really optimise, as they really weren't built originally with physics in mind.

Then of course we work with publishers and hardware vendors to get our product out there. We're picking up major traction, and the reason for that is we're starting to see some games coming out based on the work we've done with developers.

We make our money based on hardware, in the PC space, based on chips that we sell. But, for developers, we needed to make sure our software was cross-platform – you can take our software and compile it for the right platform, be it PS3, 360, Wii or PC. Actually, right now, there's at least one game on every console right now that uses our software.

You can count on one hand the engines that really matter these days, and one of the most important of those is UE3. We've got a good relationship with Epic, and we've made specific optimisations into the UE3 engine making it not only perform better but also easier for developers to use. So we've knocked out some bottlenecks that existed – there were some bottlenecks on the CPU that prevented the physics processor being used that well.

We've worked with NetDevil to do two things: one being to optimise the UE3 engine and then taking those optimisations and rolling them into a game called Warmonger, a free five-level mini-game. What's nice about that, though, is that we can take these optimisations and roll them back into the Unreal Developer Network – and they can take this and build on top of it for future games.

We've done a lot of work with a lot of developers and put a lot of effort into our software - I think our software's changed a lot over the past year. Our hardware hasn't changed, it's still in good shape, but we've improved our SDK and we've done work with the Unreal engine.

There's a bit of a creative learning curve on the developers' side - they've never really thought that much about using physics, and when you put physics into games it changes things, right? It changes the game map, it changes your options, it changes the way you can do things.

What caused you to release the PhysX SDK for free?

In the market today it's not so much that gamers don't have a piece of hardware, it's that the PhysX technology isn't built into the games they care about in a meaningful way. So, what we see as a very important strategic step is to make the PhysX SDK and software as widespread as we can, so that we can show as many developers as possible what we can do with the SDK. So the more people we can get building with the SDK, even if it's not taking advantage of the hardware, the more people that become familiar with our technology.

Everything we do now is focused on the support space, but there are a large amount of developers out there who we don't work with but they're doing development outside of the game space. We give away our SDK for free - we've had guys come up to us at shows and they've said "hey, we're doing this crazy application we'd like to show to you" and it's something totally off-the-wall - the thing is that there's so much creativity out there in the development space that if you give them the tools to do it, you never know what will come out of it. We've got 10,000 registered and active developers on our website right now - there's a lot of activity.

And the PhysX SDK is integrated into the PlayStation 3 SDK, right?

The main reason we did that is so that the SDK is cross-platform so that developers can make the game one time and spread that technology across as many platforms as they want, because you rarely see a game that comes out on one platform these days – it might not be simultaneously, but games are usually on more than one platform. But if we don't make the SDK cross-platform, it's just exponentially increasing the effort required by the developer.

A by-product of that, though, is that - say someone builds a game on PS3. When that developer decides to bring that game to the PC, then it's just much easier for them to roll our physics out with it as well, and then they've got the foundation for having hardware support, which creates an opportunity for us.

How did your relationship with Epic come about?

Epic was one of our original partners - we've been around for about five years and Epic was one of the first companies we approached, because we saw that engine as very important and a very strategic engine in the marketplace, we knew it was going to have a lot of legs, we knew it was going to be important in the future. They were also looking at physics as a strategic move forward and they had to make a decision as to what was the best technology for it.

What we're seeing now too is that with this relationship we're now seeing a point where it's really paying off - not just for us, but also for the developers who are going to see some pretty interesting tools coming from us and Epic that support physics.

The other thing you have to think about is that a good source of creativity and innovation comes from the players themselves, from the modders - just look at Counter-Strike. Look at UE3 - it's a very moddable engine, so having PhysX in there is enticing for independent developers who want to make a name for themselves and do something cool.

UE3 is a pretty powerful tool for that, so by offering PhysX with that it's opening a lot of opportunities, and that's good for us because if there's lots of good content out there for PhysX then it spurs our business on too.

Who do you view as your main competitor - would it be Havok? What do you provide that your competitors don't?

We get asked this a lot, but I don't see Havok as a competitor. They make expensive software which is pretty good and from a feature perspective we're very competitive with our software. The difference, though, is very different - we provide the opportunity to accelerate on hardware and they don't.

12 months ago NVidia, ATi and Havok talked about a partnership to enable physics on a GPU - but there's not a single game out there that demonstrates this. I don't even think there's a demo. It doesn't surprise me, because the GPU isn't built for physics. But also, I can't imagine many developers would give up graphical power for untested physics on an architecture that's not built for it.

To directly answer your question, no, I don't think there's anyone else out there who is doing what we are - physics acceleration in hardware. You can do physics on CPU - there's no question of that - it's just a question of to what extent. Physics isn't just about a few exploding boxes and leaves flying around - physics will power things like fluids, like characters that are naturally animated, clothes, hair - the amount of processing for these sorts of things is immense, a lot more than what we see today, and there's no way that in the future CPUs will be able to take that kind of load on as well.