Nvidia helps ATI with PhysX

Nvidia know that developers will only widely adopt physics on the GPU if it's available to everyone, theres no way someone would make an AAA title like Crysis for example and put any significant effort into GPU based physics unless it ran equally well on all cards.
 
Interesting, I have to say it's kind of disappointing that ATI didn't step up to the plate. If CUDA could run on all GPUs, it could really open the door for some cool GPGPU things. I know their architecture is pretty different, but that would be really sweet to come up with GPGPU standards like this. It'd be nice to have the extra pipes that aren't really needed (I don't need over 70 fps, lol) run AI, physics, and all the other things that have been talked about, and in order for this to happen on a universal basis, standards need to be established. ATI doesn't seem to have much of an answer for CUDA, and I'd love to see this happen. It would be way cool.
 
Havok runs on the CPU. It's a totally different animal from the PhysX physics engine.
But I agree with Ferrariman, having a standard for GPGPU would allow some really cool applications and software.
 
GRAW has Havok physics engine but is physX accelerated. At least it says so on my packaging.
 
Uberbob102000 said:
Havok runs on the CPU. It's a totally different animal from the PhysX physics engine.
But I agree with Ferrariman, having a standard for GPGPU would allow some really cool applications and software.
Click to expand...
yeah but they both do physics i don't think that intel will be happy with atis desicion to use physX for physics
 
Ferrariman60 said:
I know their architecture is pretty different, but that would be really sweet to come up with GPGPU standards like this
Click to expand...
OpenCL will be supported by all.

ATI doesn't seem to have much of an answer for CUDA, and I'd love to see this happen. It would be way cool.
Click to expand...
There is CAL/Brook+. NV just threw a ton of money into marketing CUDA (i.e. the 500 million dollar 'donation' to UIUC), which is why it's more visible at the moment.

As much as I hate the idea of adopting a NV standard, it could be interesting to be the only company that can accelerate both Physx and Havok in hardware.
 
I could not give jack about Vantage score, however I do care about the matter in principal. The sad fact is that 3Dmark scores sell cards and thus is the first one to recieve some special "attention".

Then again, hardly suprising, both parties have been caught doing this in the past. Shame that Nvidia has not learned from it.
 
morfinx said:
There is CAL/Brook+. NV just threw a ton of money into marketing CUDA (i.e. the 500 million dollar 'donation' to UIUC), which is why it's more visible at the moment.
Click to expand...

CUDA is a higher level language than CAL and Brook+ has problems with applications that use it, breaking with driver revisions.

Also, UIUC was using CUDA for quite a while now. The "donation" you speak of was to acknowledge the U as a CUDA Center of Excelence.
 
Silus said:
CUDA is a higher level language than CAL and Brook+ has problems with applications that use it, breaking with driver revisions.

Also, UIUC was using CUDA for quite a while now. The "donation" you speak of was to acknowledge the U as a CUDA Center of Excelence.
Click to expand...

You make having the option of using less abstraction sound like a bad thing. Where did you see that Brook+ has problems?

NV is really throwing a ton of money at promoting CUDA. Why is NV pushing CUDA so hard? Because they have to. Graphics is not a growth area - total shipments have been flat for the last several years. With their bread and butter chipset and graphics business under increasing pressure, the only way to grow its business is to branch into or create a new market.
 
morfinx said:
You make having the option of using less abstraction sound like a bad thing. Where did you see that Brook+ has problems?
Click to expand...

It is a bad thing, if you want to increase its adoption rate. C/C++, Java especially, but any other high level programming language, are the bread and butter of most developers, in which I'm included in. Low level languages ? Not so much and you're hard pressed to find many jobs, with specific low-level language requirements.

morfinx said:
NV is really throwing a ton of money at promoting CUDA. Why is NV pushing CUDA so hard? Because they have to. Graphics is not a growth area - total shipments have been flat for the last several years. With their bread and butter chipset and graphics business under increasing pressure, the only way to grow its business is to branch into or create a new market.
Click to expand...

Why wouldn't they ? It's their product and NVIDIA is a money making company. Touting their product and making it more appealing for everyone, is what makes them a profitable company. Don't really see the surprise there.
 
You must log in or register to reply here.
Back
Top