I'm just wondering if there is a super secret handshake or a written agreement that would give Intel ammunition against nVidia in this case.
So you are saying you think AMD should hold on to their judgement from Intel so they can pay nVidia that money after they are found guilty of collusion?
No. I'm saying there might be a licensing agreement in place that gives Intel ammunition to fight off the nVidia lawsuit that nVidia brought against Intel in an attempt to annul a licensing agreement that allows Intel to use some of nVidia's GPU patents. In the short term, having Intel compete with nVidia in the GPU sector benefits AMD. In the long term, this might very well be a fatal move but AMD might not have a "long term" to think about.
nVidia does not have a plurality in the graphics market, let alone anything close to resembling a monopoly. nVidia doesn't pay OEMs not to use competitors products, they don't have the leverage for anything like what Intel is being charged with across the boards by several different government agencies.
Uhh...I never claimed they did...I don't even know where this comment came from and why it is in response to something I wrote.
AMD is actively undermining Havok now, look up Bullet physics. GPU physics under Havok is still vapourware for end user purposes, Intel almost certainly wants to keep it that way until they can compete, which won't happen until when/if Larrabee ever makes it out.
I heard something about Bullet physics but to be honest with you, I haven't paid much attention to physics acceleration at all. I think it will probably be another two years before we see something decent done with physics acceleration that truly makes us wish we had it. For now, it's a checkbox feature to me. Don't take that to mean I am against physics, just that I don't feel it's there yet.
I think this may be a case of where AMD would rather Intel nip away at nVidia for them and both AMD and Intel would then collaborate and compete.
That is collusion, I would imagine Intel would be a bit loathe to take part in that at the moment given that it would be entirely illegal due to the market presence of the two companies combined. It would be rather moronic on AMD's part to put it mildly. Really now, is having slightly slower AA performance in one game the same as having someone pay out millions of dollars to have your customers not use your products? Are people here really ignorant enough to think that one company having more active developer relations is remotely close to what Intel has been doing?
Uhh...no. It can be collusion but if done right it's just business. To a degree, AMD has a certain set market in the integrated GPU sector that is not going to change because of usage of AMD CPU's and chipsets. There's likely a bundling discount going on there. Intel has the same with their integrated GPU chipsets along with their CPU's.
What's likely going to happen if Intel truly gets the ball rolling on Larrabee is they chip away at both AMD's GPU market share as well as nVidia. But it is much more likely they chip away at nVidia by stealing away a lot of the lower end discrete GPU sales that nVidia currently enjoys. That's what I meant. It's pure speculation and there is a very high probability that they would steal a huge chunk of AMD's market share but again, my own speculation is that it is much more likely that nVidia has more to lose with a new GPU competitor.
There is no collusion if all AMD is banking on is if they help allow Intel gain a foothold that nVidia is going to be hurt more than AMD. There is only collusion if both actively agree to hurt nVidia via illegal business practices.
To a very large degree nVidia steers the direction in the GPU world. Having Intel and AMD as viable and healthy competitors, especially given Intel's integrated GPU share, means they can dictate standards and move the industry in a way that benefits them. There does not need to be any locking out of nVidia much like how AMD can license PhysX or nVidia can license Havok at any time. But much like how AMD would be a generation behind on many of Intel's new SSID enhancements to the x86 architecture, nVidia might find itself having to play catch up in many cases to Intel and AMD. This is why it is a bad idea, business wise, for AMD to license PhysX. Something the fanboys still can't wrap their heads around. It may be "dirty" to a degree but that doesn't mean it's collusion.
On die GPUs will replace integrated graphics, and that's it for the near future. We also don't know yet if Intel will be allowed to make GPUs at all after their upcoming case, we will have to wait and see the outcome of that.
Yeah. That's why I'm wondering if maybe Intel has an agreement in place or nearing one with AMD that would give it a leg up on negotiations with nVidia. Something that would give it enough ammunition so to speak to fight off nVidia enough that they can get new licensing agreements in place.
It's similar to how Intel made concessions to AMD in this case but AMD also made concessions in this new agreement. AMD probably could have gotten more money from Intel in the end but that would take years. AMD also had a very real possibility of losing their x86 license. AMD gave some ground, Intel gave some ground, agreement reached.
I'm just saying that the more widely usable standard will get used.
You mean like how OpenGL beat DirectX? How Linux beat Windows? The most widely usable standard assures you nothing. Direct Compute btw, isn't likely to be the more useable standard. It will only run on Windows platforms, OpenCL will be, but CUDA is fairly likely to be supported on more platforms then DC(portables, consoles, HPCs, and PCs).
Here's the full quote so we can set what you quoted in the proper context. "The CUDA and PhysX support is still small enough that strong efforts by AMD and Intel can weaken it to the point of making it irrelevant. Not here to argue nor concede whether CUDA is better than OpenCL & DirectCompute or whether PhysX is better than Havok. I'm just saying that the more widely usable standard will get used. Beta was better than VHS but we all know how that played out. If OpenCL & DirectCompute works on pretty much all GPU's then why support CUDA?"
I'm saying that if Intel gets a foothold in the industry then standards supported by AMD and Intel can undermine any standards set by nVidia. Such as is the case where DirectCompute is supported by AMD and Intel against CUDA which is supported and created by nVidia. And by more usable I meant more widely supported on all GPU's rather than CUDA which would only work on nVidia GPU's. Not from an elegance or ease of use standpoint but from a number of supported device standpoint.
CUDA may be supported on more non Windows devices but it's hard to argue that there isn't competition on Windows. Especially with Microsoft creating and backing a competitor to CUDA and the second largest discrete GPU maker against CUDA and likely the largest GPU vendor against CUDA as well.
Bullet is a much better option then Havok atm, they just need to get it done ASAP.
/shrug
Dunno. Don't really care. Physics acceleration as a whole doesn't interest me at all because I don't feel it is used in any game changing way at the moment. Mostly for added eye candy. When we get a game that utilizes it to create a more realistic gaming experience then I'll get interested. My own feeling is it won't be until 2011 when we get games built with physics acceleration in mind.