- Tuesday, 10 November 2009
- INQ Mobile
- RSS
Anything from yesterday is probably too old ',INQUIRER School of Journalism" - 1 "791"
Physx tumbles forward
RECENT SPECULATION about the anticipated demise of
GPU
physics and the release of UT3 has discrete physics at a crossroads.
The recent price cuts of the Ageia Physx processor has brought the add in physics module into the realm of affordability as, in general, people were more inclined to invest their hard earned money in a graphics card upgrade as they would realize a benefit in every game, not the smattering of titles currently supporting Physx.
A cynical mind might suggest that the ‘physics on GPU’ announcements by both DAAMIT as ATI and Nvidia were designed to create confusion in the developer industry. If there are several competing formats, with whom do you throw your lot?
Physx has a narrow window in which to succeed but several of the factors affecting this window are completely beyond Ageia's control. The critical part of the equation is not the hardware which is proven, but the title support.
To see the full benefit of Physx you need to build the game around the engine so that you are able to implement game play related physics as opposed to the ‘flowing goo from a barrel’ physics. Currently titles are tending towards being multi-threaded which can use the inexpensive quad core chips coming in early ’08.
Unfortunately this is still the wild west with several different
approaches to programming for multi-core CPUs. If Ageia can swing another few
franchise titles before physics moves to one of those cores it might survive as
currently structured.
If this doesn’t happen, the outfit might transition to an IP-based company by porting instructions to a CPU core - and become the Rambus of the physics world. µ
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
Looks like MS left an legacy in manipulation and domination so other companies as the graphic cards manufacturers to use there muscles to stop or make difficult for a good idea to go main stream.
There's some free game about to be launched call warmonger which will show what the ageia can do...

Totally destructable enviros...imagine shooting the floor out from under an enemy....if this game works, then it might prove their case to other developers.
Maybe what it is needed is a OS level physics instructions, like there is to graphics through Direct X. While I don't know whether that is possible, this situation sure reminds me of 3dfx "glide" vs PowerVR..

Create a high lvl PX language, drivers to convert that into the PPU native instructions, whether a "true" PPU or a GPU powered emulation PPU (or whatever) - or a CPU core emulated one.

Not perfect? 

The world seldom is...

and other completely optional add-ons, I think the PhysX chip is not going to truly be successful until it starts being bundled on other PCBs.
I always wondered why did they even bother to make their IP into a separate PCI card. it just screams to be integrated into existing GPU designs or even as part of the video card hardware.

That would have had eliminated so many of the hurdles they are facing now...

This is such an obvious approach that I think they probably:
- Got greedy and wanted it all
- Got greedy and wanted too much money for their license

They way things are going, I see them getting bought up by Intel after going down.
As stated game support isn't exactly Ageia's fault, quad cores are suffering from the exact same problem of getting the install base needed to justify making it a mandatory requirement. You don't make games only a fraction of people can play, but that install base doesn't go up without justification to upgrade. 

So the only way to get around the problem is to put money down yourself and make some compatible games, like Cellfactor and Warmonger. The problem however is actually making a game and not a physics tech demo, neither of those games have the substance needed to keep users interested. 

I think Ageia's best bet is with modular games like UT3; that way you can go all out with the physics game play without needing the game itself to fully utilise it, they also don't need much depth so it avoids Ageia's game design weakness.
The Physx software already does work fine without the hardware. Obviously it's not as fast that way, and it's not clear how well it would compete on either features or performance with other software-only solutions.
I don't know for sure, but as far as I can see, GPGPU physics requires 2 or 3 graphics cards.

I already do not buy into SLI because I find the return on investment to be ridiculously small, so buying 3 of them is so far out of the question that it's not even funny anymore.

I find a company pushing me to buy 3 of the same to be a lot more greedy than a company proposing an add-in PCI or PCIe card. Maybe that's just me.