• 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.

Nvidia/Gpu/Physx vs dedicated Physx card?

ziscwg

Limp Gawd
Joined
Feb 20, 2004
Messages
354
Any difference?

I only have a 1x (PCIE) slot and PCI slots open. (only 1 16x slot on my board)

So, if I went Physx, it would have to be a dedicated card. I still will be waiting to see more compelling content, but I'm actually thinking about getting one now.

With Nvidia releasing some drivers to let you run Physx on a newer GPU, I'm hoping developers really start using it.
 
a ppu is only a little slower than using a 9600gt for physx from what i've read. in your case, you might be better off in the long run just investing a bit more by upgrading your video card to like a 9800gtx to run graphics + physx if it fits in your budget.
 
I think the only reason to go dedicated physix is if your current card doesnt have it, if there are current products available (which there is) that has it integrated im sure it would do the job just fine.
 
Very few of the benchmarks that have been released since the latest PhysX drivers came out include figures from the old PCI PPU card so it’s still hard to get a clear picture of where things stand at this point.

A few things to consider:

The OP said that he only has PCI and PCIe 1x slots available. Now obviously there are PCI and PCIe 1x PPU cards available, but as far as cards that can emulate PhysX via Cuda, there is a PCI version of the 8500 that is sold a few places though as far as I know no one has done any benchmarks using the PCI version so no one really knows if it would be faster or if it would even work at all. Perhaps there is additional overhead associated with Cuda emulation that makes the limited PCI bandwidth more of an issue than it was with the PPU. I’d certainly be hesitant until we see some numbers.

There are also power and driver issues to consider. There are known issues with Vista right now which require you to have a 2nd card hooked up to a monitor in order to utilize it for PhysX; this limitation is obviously not in place for a PPU. Also, the PPU represents a paltry 28w TDP which should be a bit less than any secondary GPU you might put in there just for PhysX, thus saving on power and heat.

Finally, a PPU gives you a certain degree of additional freedom in choosing your primary videocard. Especially in Vista, which doesn’t play very well with video drivers from multiple companies installed at the same time, having a PPU allows you the choice of buying an ATi card while continuing to benefit from PhysX. I’m currently running a 4850 along side my PPU. Even if the PPU is slower than Cuda emulation, it’s certainly a lot faster in PhysX games than no acceleration at all, and the additional freedom in choosing your primary videocard is definitely worth something. I was getting just as good if not better performance with my 4850+PPU in UT3 on PhysX levels than I was getting before just using my 8800GT.
 
Rationalize your ppu purchase all you want. The fact remains 8xxx gpu >> ppu with Physx.
 
The OP said that he only has PCI and PCIe 1x slots available. Now obviously there are PCI and PCIe 1x PPU cards available, but as far as cards that can emulate PhysX via Cuda, there is a PCI version of the 8500 that is sold a few places though as far as I know no one has done any benchmarks using the PCI version so no one really knows if it would be faster or if it would even work at all. Perhaps there is additional overhead associated with Cuda emulation that makes the limited PCI bandwidth more of an issue than it was with the PPU. I’d certainly be hesitant until we see some numbers.

There isn't. Physics aren't bandwidth-intensive at all, at least not for the memory and bus.
All the heavy lifting happens inside the chip. The actual physics data is very small. It's more or less geometry data, usually a very simplified version of the visual geometry.
I'd be surprised if you'd notice any difference in performance at all between PCI and PCIe for this purpose.
 
Rationalize your ppu purchase all you want. The fact remains 8xxx gpu >> ppu with Physx.

And when someone does benchmarks of the 8500GT PCI, currently the only card which could potentially run PhysX that fits the OP's situation, then your comment might become relevant. There are still a lot of assumptions being made that it would even work, etc, vs the PPU that is a known piece of hardware for exactly this type of purpose. Also FWIW I got mine for free nearly two and a half years ago so I'm hardly trying to justify my purchase. :rolleyes:
 
Rationalize your ppu purchase all you want. The fact remains 8xxx gpu >> ppu with Physx.

perception (opinion) and reality (fact) are two different things. and the 8xxx gpu part is rather vague. i really don't see an 8400gs improving hardware physx performance.
 
Back
Top