• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Does Nvidia PhysX require an SLI board?

S

skinnypotbelly

Limp Gawd
Joined
Aug 14, 2005
Messages
371
Does Nvidia PhysX require a board to be SLI capable? Or can a intel chipset with the necessary pcie slots run PhysX also? IE will my Maximus formula be able to run PhysX? Or do I need to switch to a 780i, 790i, etc to run it?
 
They haven't said exactly how it will work, but I really doubt it. They are marketing this as a feature on their cards, not on their motherboards.
 
I believe that it will work like CUDA, so i think that it will be able to process Graphics and Physics but if you have SLi you will get better graphics and physics, and adding more GPUs probably means more performance with both.
 
RussianHAXOR said:
I believe that it will work like CUDA, so i think that it will be able to process Graphics and Physics but if you have SLi you will get better graphics and physics, and adding more GPUs probably means more performance with both.
Click to expand...

Sorry, but the card can only do graphics or run CUDA...not both at the same time.
It's not like graphics and physics will run seemlessly side-by-side on a single card.
 
Sorry, but the card can only do graphics or run CUDA...not both at the same time.
It's not like graphics and physics will run seemlessly side-by-side on a single card.
Click to expand...

Yes you can, matter of fact you can do it already, just go download CUDA SDK which is free and run through the tutorial, setup a routine and run a game at the same time, it simply just works.

What you cannot do at this time is use mipmapped textures and anisotropic filtering within CUDA, but this is not the case, the games are not built to run inside CUDA but to access the API for physic calculation, the same way you would run a routine.

It's not any harder than running 2 process on the same CPU (solo core for example)Furthermore the stream processor can access the motherboard memory instead of videocard memory using the texture memory which won't affect the videomemory space for high resolutions.or simply use the gpu onboard memory as shown in this link.
http://www.biomedcentral.com/1471-2105/8/474/figure/F2
 
The way things run now is like this:
"Interleaved kernels" which get 100% of card resources
This is what happens now with CUDA on a graphics device. The CUDA kernel owns the card during execution, and the graphics driver does updates between calls.

But lets go back to this when they actuall release something else than PR, and I will bet that they are aiming for multiple card solutions...
 
The way things run now is like this:
"Interleaved kernels" which get 100% of card resources
This is what happens now with CUDA on a graphics device. The CUDA kernel owns the card during execution, and the graphics driver does updates between calls.
Click to expand...
Which produce 1ms blips, but still you can use the same card for video, thing is, and that's what I find disturbing is that nVidia recommend one CPU core per GPU under CUDA to alleviate the kernel loading times.

Anyway, all we have so far are technical demo's and PR, btw did you guys know you could plug 2 ro more physx card with drivers 2.7 and up and they would be recognise by the drivers?(previous versions wouldn't recognise more than one)
 
You must log in or register to reply here.
Back
Top