Not sure what software you are referring to there
Here's a nice recent (and current) example:
AMD tout
VCE as a significant part of their GCN architecture...yet have no functioning drivers for the feature. Dev's aren't coding for VCE (maybe ArcSoft?) in part because they have no functioning driver available for Radeon HD 7000 series with which to run/test/implement the feature....and AMD probably wont push implementing the feature until the software library becomes available. Familiar with the phrase "Catch-22"?
Easy enough to find similar -and more glaring problems- with the late and constantly changing roadmaps associated with AMD's Stream (
now APP). If you're looking solely at gaming -a rather narrower emphasis- then you still see less-than-shining examples with AMD's loss of interest in GITG, recent driver woes in Crossfire/GCN, and exactly how long was it between HD 7970/7950 "launch" on 22nd December to the appearance of WHQL drivers ?
...and the argument goes that games developers would benefit more from the competition of the GPU market.
Sounds like an argument worthy of a special needs institution.
Dev's benefit from sponsorship by companies via SDK's or direct coding or wads of cash.
Gaming is not dependant upon competition in the GPU market- at least not directly. AMD and Nvidia haven't gone seriously head-to-head in a price war since 2007-08. A few minor skirmishes ( GTX 460 v HD 5830, GTX 260 v HD 4870 etc.) aside, they have managed by some minor miracle-
or otherwise- to dovetail price and performance-the exceptions being the top-dog SKU of any generation. All the present/future pricing structure does is to limit uptake of new parts if they are 1. too expensive for the performance gained, and 2. the Osborne effect comes into play if a newer series is imminent
To allow this to be a more level playing field, an open physics standard would be beneficial.
There are already ample iterations of physics engines in gaming aside from PhysX. Bullet, Havok (Intel) and a whole raft of game specific physics applications.
The fact also remains that proprietry or company driven standards develop and mature faster than open source as a general rule. Don't agree? then look at the development cycle of OpenGL and lately, OpenCL.
AMD obviously agrees with this because it helps them gain market share as their products are more on par with the competition wrt features.
You know many people that base a graphics card purchase primarily on it's ability to use game physics?
NVIDIA doesn't because they have the market share to lose and are happy with the situation at the moment.
Marketing bulletpoint only- As are the majority of in-game physics effects.
With all the anti-nvidia rhetoric I still haven't heard who has the job of debuggiing and code implementation for incorporating into AMD drivers. Would you see this -hypothetically-as something the Nvidia driver team should pick up the tab for ?
TL;DR : There's no way on Earth that AMD would allow Nvidia anywhere near their driver code, and AMD have zero interest in PhysX in any case.
Game developers have the choice of what game engine they utlize, and what-if any- physics ettects they incorporate. For the smaller game studio's maybe the incentive is there to add a sponsors codepath...so what's the alternative? No SDK and development help, and late release (or not at all)?...Open source game or no game? I just love the pontificating from on high regarding open standards and the advancement for gaming when AMD folded their tent for
5+ years as far as providing game devs with sponsored support.
Ever wonder why Nvidia has such entrenched gamer support even in the face of generally more expensive cards that frequently arrive late and use more power? Maybe it's a continued presence within gaming- call it mind share. You can call it marketing, pushing a proprietry standard or any number of other money-speak catch phrases...it still amounts to Nvidia putting funding into gaming when basically no one else was interested.
PhysX is nothing more than an added level of eye-candy. It changes nothing regarding gameplay. Nvidia's cards' CUDA ability also allows for shader effects such as bokeh filtering- it's an optional extra if you have a card that can process the workload whilst still maintaining a playable framerate -some people would see that as added-value feature, just as Nvidia has offered transparency antialiasing.
So long as the additions
are additions and not handicaps for the opposing teams consumers (ie the AA issue with Batman:AA) then I have no problem with implementing the effects- proprietry or not- and if it means the difference between a game being launched or not then that should be a non-brainer.