Ageia Sold

[Dirk Gregorius]

I just read that Ageia was sold as well. Unfornutately they author doesn't know the new owner. It seems not to be AMD and it is also not clear if Intel or nVidia bought Ageia.

http://www.fudzilla.com/index.php?optio ... 0&Itemid=1


Edit:
It just seemed to be a rumor:

http://www.pcgameshardware.de/?article_id=629108

The article basically reads that the connected Ageia and they denied the aquisition.

[KenB]

Imagine Intel buying Ageia, and thus owning both Havok and Ageia.
Interesting situation. Endless discussions about everything to be done.
Who would like to manage that?

Anyone heard anything about a PhysX generation two card?
The original card har been around forever (if time is measured in terms of
how e.g. graphics cards have developed).

I'm not attempting to start rumours. I'm just curious whether there is
a roadmap from Ageia that has been presented somewhere.
Most other hardware vendors do have fairly official plans for their
upcoming architectures.

[Dirk Gregorius]

>>Imagine Intel buying Ageia, and thus owning both Havok and Ageia.
Intel will own a lot of patents then. (Prior art or not. Rights and wrongs about patents also not important for a moment). I think nobody wants a lawsuit with Intel about violating one of their patents.


With respect to Ageia I haven't heared anything about a next generation card. Actually I only heard that some people left (e.g. Pierre Terdiman and Dennis ? from Meqon). My assumption is that their business plane was to be bought by someone. The problem is that they are not worth 100M like Havok in my opinion and I think the venture capital was around this much. So maybe Intel buys them if the price is right (read cheap). AMD seems to struggle with ATI, so it is questionable if they could afford to buy them.

To my knowledge the physic card had no special inventions and so why should nVidia be interested. Actually we also don't know how much know-how nVidia gained from their work with Havok on the GPU-FX engine.

Most probably much ado about nothing in my opinion and nobody will talk about Ageia in one or two years anymore.

[Pierre]

Mathengine... Ipion... NovodeX... Meqon... Havok... Ageia (*)...

All of them acquired. Physics engine business sucks If I would have a game company starting a new project, I'm not sure what physics solution I would end up using those days.

- Pierre

(*) I left indeed so I have no idea if the rumors are true.

[raigan2]

I thought the whole point of any of these companies (or game companies in general) was to be bought for huge amounts?!

[Dirk Gregorius]

How many choices you actually have. I would only choose between Ageia or Havok. Also what is being bad about be bought for a huge amount of money. I would be drunk one week if my company get aquired by e.g. Microsoft

[Erwin Coumans]

How many choices you actually have. I would only choose between Ageia or Havok. Also what is being bad about be bought for a huge amount of money. I would be drunk one week if my company get aquired by e.g. Microsoft

Why ignore open source physics engines like Open Dynamics Engine or Bullet?

Many game companies prefer open source or inhouse over PhysX or Havok. They ofen take those open source engines as a starting point and continue optimizing and improving from there. That way you can keep full control and freedom to customize any part of the source code. It seems that the gap in performance is closing too.

Thanks,
Erwin

[Antonio Martini]

i believe that the choice is dictated by many factors and among others:

- available time.
- available resources and expertise
- company short/medium/long term plans
- required physics quality over time
-....

cheers,
Antonio

[aboeing]

Mathengine... Ipion... NovodeX... Meqon... Havok... Ageia (*)...

All of them acquired. Physics engine business sucks If I would have a game company starting a new project, I'm not sure what physics solution I would end up using those days.

Perhaps a physics engine abstraction layer would come in handy

Why ignore open source physics engines like Open Dynamics Engine or Bullet?

In my experience it seems to be a perception of open source = worse quality from programmers, and for management open source = no support. It would seem the first perception is false, whilst the second is more accurate. Perhaps sourceforges services system will solve that problem eventually...

It seems that the gap in performance is closing too.

Non-commercial/"research" engines outperform the commercial ones in my experience. The problem is, non-commercial or "research" engines are just made for one specific task, and are not generalizable to the problems most people want to solve with the physics system.

[Pierre]

Perhaps a physics engine abstraction layer would come in handy

Not really Integrating X or Y is not really the issue, it's more a question of betting on the right horse - one that won't die or collapse before the end of the race, leaving you with no support . Adding library X or Y is usually not the problem, most of the time there's already an in-house wrapper around the physics library, and switching from one to the other is just a matter of changing the wrapper's implementation (usually one file) - not its interface. The problems start afterwards when library X doesn't behave exactly like library Y, and you have to re-tweak all your parameters. (I don't know if PAL helps for this.)

open source = worse quality from programmers

It's more like "open source = variable or random quality from programmers". You can get jewels and pieces of crap in the same file sometimes.

open source = no support

That's more like it. And also the open source reaction time for painful-things-to-implement is huge. If nobody's "interested" in a feature, nobody will implement it for you. Another problem is that open source without a leader sometimes gives birth to endless discussions about recurrent, pointless topics - basically nobody ever agrees on what's important.

Non-commercial/"research" engines outperform the commercial ones in my experience.

Any example?

- Pierre

[Dirk Gregorius]

Why ignore open source physics engines like Open Dynamics Engine or Bullet?

I agree with you argumentation. If I needed to write an inhouse engine and had no prior experience I would definitely start from an open source implementation and contribute back within the rules of the NDA. Still there are arguments that speak for commercial engines and Pierre made some valid points.

[Erwin Coumans]

The discussion about a potential Ageia sale, and the software relationship to alternatives (like Havok, open source, in-house) is very important and central to this forum. For the discussion, let's split the Ageia properties into PhysX hardware and the Novodex physics software.

• A physics engine is a very good showcase for gaming hardware: Intel or AMD multi-core CPUs like Larrabee, IBM/Sony/Toshiba Cell SPUs or NVidia/AMD GPUs. This dependency is quite interesting, and chip manufacturers all want their chips to perform best.
• Ageia PhysX card competes with every hardware manufacturer, and I don't see much intellectual property in the Ageia hardware of interest to big hardware manufacturers.
• The Novodex physics library competes with Havok, ODE, Bullet etc.
• The Intel purchase made Havok weaker in my opinion, because game companies and publishers will unavoidably doubt that Havok will put enough effort on competing platforms.
• In that perspective, I think open source is a better business model for physics engines, especially with respect to multi-platform. It is more independent from hardware manufacturers.

switching from one to the other is just a matter of changing the wrapper's implementation (usually one file) - not its interface. The problems start afterwards when library X doesn't behave exactly like library Y, and you have to re-tweak all your parameters. (I don't know if PAL helps for this.)

Agreed, for the basic setup you can have a wrapper, but tuning all the settings is a lot of work. Nevertheless NOT having such wrapper demotivates people to try out alternatives. Many game companies have such physics abstraction layer, but they usually don't have the time/resources/interest to implement it for yet another physics engine. If that implementation is readily available (through PAL etc) then it makes it easier to compare engines at the start of a new game project.

It's more like "open source = variable or random quality from programmers". You can get jewels and pieces of crap in the same file sometimes.

Indeed, but on the flipside, closed source physics engines can be much harder (or even impossible) to customize as you like.

open source = no support

That's more like it. And also the open source reaction time for painful-things-to-implement is huge. If nobody's "interested" in a feature, nobody will implement it for you. Another problem is that open source without a leader sometimes gives birth to endless discussions about recurrent, pointless topics - basically nobody ever agrees on what's important.

Indeed, just like closed source, open source also needs organization and a motivated leadership, also called benevolent dictator. For Bullet that is me obviously Ideally you have one or more big companies employing the developers of open source with mutual benefits. Open source can have very good support through forums like this.

Non-commercial/"research" engines outperform the commercial ones in my experience.

Actual performance and quality comparison is an interesting issue which deserves a thread of its own, better to continue discussing comparison details there.

A few more points I want to make:

• The development of a (closed and open source) physics engine should be driven by its customers in my opinion. In the case of Bullet, the boundary between user and developer is blurred: several commercial game companies contribute changes back to the project.
• Closed source physics engines carry the risk that the company goes out of business, and the software becomes unavailable. Remember Criterion Renderware? Bought up by EA, and then many game companies and publishers were out of luck... This doesn't happen to open source. Once the software is out, you are free to take it, modify it and no-one takes it away from you.
• The book Producing Open Source Software, goes into detail on proprietary versus open source. It has been written by the author of Subversion. The online version and the pdf of the book is a free download that I highly recommend.

Erwin