How much did PhysX support factor into your purchasing decision?

How much did PhysX support factor into your purchasing decision?


  • Total voters
    171
PhysX support factored in more than "very little," but less than "heavily." It's nice to have when it costs nothing extra, even with the relatively low number of game supported. CUDA and 64-bit FP support factored in more for my last purchase (GTX 260 Core 216).
 
I give up, again you don't seem interested in the politics of it, there is no point forcing something if it is not in the best interested of the public.

You or others can give other examples of when it has happened but the fact that it has happened before does not make it right,

This whole industry needs to grow up and mature to a point where it is done to drive things forward, not all the destructive competition. Competition should be used to drive things forward not create a stalemate.

Aren't we the idealist :)

Fact is that Havok isn't going anywhere any time soon (blame Intel). AMD isn't going to mess with a physics API from the looks of it. Only nVidia has an API and wants to allow everyone to get the most out of it.

Perhaps you're willing to wait a few years for things to change and a completely open physics API standard to pop up, but I bet there are many more who are more than happy with what nVidia is offering right now. I know I am.

I wasn't going to buy an AMD/ATI card anyway, as the OpenGL support on those cards is inferior to the implementation on nVidia cards. How is that for supporting open standards? ATi has been hugging really close to a proprietary, closed 'standard' like DX for decades now. For them to whine about open standards now is hypocritical to no end.
 
PhysX wasn't around when I bought my card so it wasn't a factor at all. But it's nice to be able to play with new tech, it's one of the little extras that make your ownership experience a little more fulfilling. It's not much but it beats reading about stuff that I can't play with first hand.

Oh, by the way the poll options are skewed towards either being a PhysX fanatic or dismissing it outright. Where's the middle ground?
 
I'm not worried about how realistic barrels bounce off each other. I'm playing games to have fun and my 4870 helps me do so. So no Physx support played a 0% role in my purchase.
 
physx, not so much, but cuda, yea.
a pretty well developed language like that could potentially be invaluable, giving nongaming uses for graphics cards
 
I mean nvidia has alot to gain from public acceptance but how do the public really benefit from a company like nvidia gaining success in this field.

The same way the public has benefited from the successes of IBM, Intel, Microsoft, Google. I'm sure you can make the necessary connections.

And exactly what do you mean by "a company like Nvidia"? Would a company like AMD or a company like Intel having success be better for all of us? I'd love to hear the argument behind that particular stance.
 
physx, not so much, but cuda, yea.
a pretty well developed language like that could potentially be invaluable, giving nongaming uses for graphics cards

Agreed. CUDA is invaluable to my line of work as well (GPGPU applications). OpenCL will still take until 2010 to pan out (especially for AMD...), so to me nVidia is the right way to go :)
 
OpenCL will still take until 2010 to pan out (especially for AMD...)

What makes you say that? nVidia has said they will have beta drivers with OpenCL support in Q1'09, and a release in Q2.
AMD seemed to be more directly involved with the OpenCL standard than nVidia, so I assume they will have support at the same time, if not sooner.
I assume Intel will come out of of the gate with OpenCL on Larrabee. They will probably have OpenCL support on their regular CPUs even sooner.
 
AMD seemed to be more directly involved with the OpenCL standard than nVidia, so I assume they will have support at the same time, if not sooner.

The Chairman of the Khronos group that developed the standard is an Nvidia VP, formerly Ageia's CEO. What makes you think AMD was more directly involved? OpenCL turned out looking more similiar to CUDA than anything else out there. You sure you're not thinking about Apple who first proposed that such a standard be developed?
 
What makes you say that? nVidia has said they will have beta drivers with OpenCL support in Q1'09, and a release in Q2.
AMD seemed to be more directly involved with the OpenCL standard than nVidia, so I assume they will have support at the same time, if not sooner.
I assume Intel will come out of of the gate with OpenCL on Larrabee. They will probably have OpenCL support on their regular CPUs even sooner.

NVidia said H1'09, yes. AMD, however, said H2'09. Effectively AMD will be releasing its implementation of OCL in Q4'09. I have heard nothing on Intel yet.
 
The Chairman of the Khronos group that developed the standard is an Nvidia VP, formerly Ageia's CEO. What makes you think AMD was more directly involved? OpenCL turned out looking more similiar to CUDA than anything else out there. You sure you're not thinking about Apple who first proposed that such a standard be developed?

I know it's Apple who went to Khronos for a standard (just tough luck for AMD that there happened to an nVidia chairman... but I don't think that influenced the situation much, if at all)... nVidia already has Cuda on the market, and indeed, OpenCL is basically a clone of Cuda (just like OpenGL is now more or less a clone of Direct3D, their shader language is almost exactly the same as Cg and D3D's HLSL).
For nVidia OpenCL isn't half as relevant as it is for AMD. nVidia already has Cuda, which is very similar.
Until recently, AMD only had the very poor Close-To-Metal thing, with Brook+ bolted on top. They recently started over with their Stream SDK, which no doubt was built on early drafts of the OpenCL standard, and will support the official OpenCL 1.0 standard in the coming 1.4 release.
 
Aren't we the idealist :)

Fact is that Havok isn't going anywhere any time soon (blame Intel). AMD isn't going to mess with a physics API from the looks of it. Only nVidia has an API and wants to allow everyone to get the most out of it.

Perhaps you're willing to wait a few years for things to change and a completely open physics API standard to pop up, but I bet there are many more who are more than happy with what nVidia is offering right now. I know I am.

I wasn't going to buy an AMD/ATI card anyway, as the OpenGL support on those cards is inferior to the implementation on nVidia cards. How is that for supporting open standards? ATi has been hugging really close to a proprietary, closed 'standard' like DX for decades now. For them to whine about open standards now is hypocritical to no end.


Considering how poorly nV's top of the line cards perform with Cryostasis I see no value in this at all.
 
For nVidia OpenCL isn't half as relevant as it is for AMD. nVidia already has Cuda, which is very similar.

Oh, but it's even more relevant simply for that reason. Nvidia spent all this time and money pushing GPGPU and developing and evangelizing CUDA to the masses. They're not about to let an open standard steal their thunder. They're going to put their full weight behind OpenCL to ensure that they retain their leadership in the field. In a faceoff OpenCL beats CUDA simply because it's an open cross-platform standard so Nvidia isn't going to sit idly be and be marginalized. They want to be known as the kings of GPGPU.
 
Oh, but it's even more relevant simply for that reason. Nvidia spent all this time and money pushing GPGPU and developing and evangelizing CUDA to the masses. They're not about to let an open standard steal their thunder. They're going to put their full weight behind OpenCL to ensure that they retain their leadership in the field. In a faceoff OpenCL beats CUDA simply because it's an open cross-platform standard so Nvidia isn't going to sit idly be and be marginalized. They want to be known as the kings of GPGPU.

I don't think you get what I mean.
OpenCL is almost identical to Cuda. This means that:
1) It won't take nVidia a lot of effort to take their Cuda SDK and modify it to support OpenCL. AMD on the other hand had to start over.
2) Since it's so similar to Cuda, and Cuda is specially designed to take advantage of nVidia's hardware architecture, OpenCL will also run very efficiently on nVidia's architecture. Since AMD's architecture is quite different, OpenCL might not get the most out of it... so they'll have to start over on the hardware aswell.
3) There already are a lot of developers working with Cuda. Even if they move to OpenCL, they will likely continue to use nVidia products, because that's what they're familiar with, and nVidia has left a great impression with Cuda. AMD has no developer base at this point.

So basically OpenCL hands nVidia the GPGPU market on a silver platter.
 
..., OpenCL is basically a clone of Cuda (just like OpenGL is now more or less a clone of Direct3D, their shader language is almost exactly the same as Cg and D3D's HLSL).

Considering how DX is very static and OGL is the one always progressing thanks to extensions, I'd say that D3D is borrowing from OGL :p
 
Considering how DX is very static and OGL is the one always progressing thanks to extensions, I'd say that D3D is borrowing from OGL :p

Firstly, I don't consider proprietary extensions to a standard a part of the standard itself.
After all, they don't work on other standard-compliant hardware, since they don't support those proprietary extensions.
Once they do standardize extensions (such as GLSL or vertexbuffers or such), they borrow them from D3D.

Secondly, since Direct3D 9 it's the API design that drives the hardware, not the other way around. Microsoft and the various GPU manufacturers first standardize the API, and then the hardware is designed. Which is why OpenGL extensions such as geometry shaders are based on Direct3D... after all, the D3D standard existed first, and only then the hardware arrived, and finally the OpenGL extensions.
You'll notice that neither nVidia nor AMD add features to their hardware between D3D versions anymore. Those days are over.
 
Which is why OGL 2.0 cards support virtually all the features of OGL 2.1 already, and why most OGL 3.0 features can be found in OGL 2.1 cards. Gotcha :)
 
Which is why OGL 2.0 cards support virtually all the features of OGL 2.1 already, and why most OGL 3.0 features can be found in OGL 2.1 cards. Gotcha :)

No I don't think you got it.
All OpenGL 3.0 features and whatever AMD/nVidia extensions are already present in DX10/DX10.1.
They weren't in OpenGL first.
 
Firstly, I don't consider proprietary extensions to a standard a part of the standard itself.
After all, they don't work on other standard-compliant hardware, since they don't support those proprietary extensions.
Once they do standardize extensions (such as GLSL or vertexbuffers or such), they borrow them from D3D.

Secondly, since Direct3D 9 it's the API design that drives the hardware, not the other way around. Microsoft and the various GPU manufacturers first standardize the API, and then the hardware is designed. Which is why OpenGL extensions such as geometry shaders are based on Direct3D... after all, the D3D standard existed first, and only then the hardware arrived, and finally the OpenGL extensions.
You'll notice that neither nVidia nor AMD add features to their hardware between D3D versions anymore. Those days are over.

Thirdly, WTF happened to staying on topic?? Here's an idea, start a new thread about it instead of hijacking this one.
 
I think we that have a physics processor subforum here, I think that it would be a good idea to keep the video card forum free of the physics discussion and use the subforum instead. I know that now people can use a graphic card to calculate physics but using a graphic card to calculate physics would mean that the card is now also a physics processor.
 
I think we that have a physics processor subforum here, I think that it would be a good idea to keep the video card forum free of the physics discussion and use the subforum instead. I know that now people can use a graphic card to calculate physics but using a graphic card to calculate physics would mean that the card is now also a physics processor.

with all due respect, i think this is pretty relevant, considering the topic of the thread deals with how physx may or may not play a role into someone's purchasing decision for a video card, which we are discussing. physx is a benefit to nvidia video cards, so nvidia is heavily marketing it as a feature for prospective buyers. and you might want to tell kyle that as well, regarding keeping physics out of this forum, considering he posted the following thread recently, lol.

http://www.hardforum.com/showthread.php?t=1373002
 
As long as the technology has brand association, it will get nowhere.

It was not a factor in my purchase.
 
with all due respect, i think this is pretty relevant, considering the topic of the thread deals with how physx may or may not play a role into someone's purchasing decision for a video card, which we are discussing. physx is a benefit to nvidia video cards, so nvidia is heavily marketing it as a feature for prospective buyers. and you might want to tell kyle that as well, regarding keeping physics out of this forum, considering he posted the following thread recently, lol.

http://www.hardforum.com/showthread.php?t=1373002

This thread is indeed relevant to this subforum but the OpenCL discussion and perhaps the thread by Kyle is better off in the PPU subforum considering that there will be no more PPU after this. The subforum will be forgotten soon and right now less than 10 people are reading the PPU subforum at one time compared to 100+ people reading the video cards subforum. My 2 cents.
 
Back
Top