• 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.

how to set up dedicated PPU?

M

MrLoLeR

Limp Gawd
Joined
Jun 28, 2008
Messages
190
I don't necessarily plan on doing this now, because i don't think i need to, but just for my curiosity, How would i setup my 8600gt to run as a dedicated PPU along side my GTX 260 as the main video card. I know i don't need an SLI capable board but what do i need to do to do that.

Just pop the card in the second PCIe slot and magically it works (i'm sure thats not how)?

Thanks guys.
 
just make sure you have the latest nvidia drivers and enable physx in the nvidia control panel, selecting the 8600gt as the physx processor. that should be it, i believe. though to be honest, i think you're better off leaving the physx processing on the gtx260, since the 8600gt won't fare very well as a "ppu".
 
Ok. I figured as much too but needed reassurance. I figured it could only lighted the load of the 260 but if you think i'd get better performance not even fooling with it then thats what i'll do.
 
Well, an 8600 GT might be on the light side for PhysX work. Recommended is an 8800 GT card or better.

Of course, you can always try with and without the 8600 GT as dedicated PPU and see how it goes :)
 
It should be, as the heavy part of physcis are the calculations, not the results.
 
Can't say for sure, but I'm guessing that the BW provided by an x1 slot will be right on the edge of what's practical. V2.0 slot would help a lot :)
 
Elledan said:
Can't say for sure, but I'm guessing that the BW provided by an x1 slot will be right on the edge of what's practical. V2.0 slot would help a lot :)
Click to expand...

According to AGEIA even PCI was fine and I see NO preformance difference betwen my PCI and my PCI-E PPU.
(Suggestion that bandwith is not a bottleneck on even PCI)
Again the x,y,z coordinates are NOT the heavy load, the only thing that is heavy about physcis are the calculations them selfs, not the bandwith (infact the lantency is more vital than bandwith)
 
Well, I'd rather trust some benchmark numbers from an independent party, but you may be right :)

And yes, latency kills. Hence why PCI PPUs are a bad idea, no matter what people may say :)
 
Elledan said:
Well, I'd rather trust some benchmark numbers from an independent party, but you may be right :)

And yes, latency kills. Hence why PCI PPUs are a bad idea, no matter what people may say :)
Click to expand...

The lantency is actuall lower on PCI than on PCI-E *hint-hint*
 
Elledan said:
Not if you're running a RAID controller/GB NIC/both on the PCI bus :p
Click to expand...

I run quite a few things on the PCI-bus and that has never reulted in lower performance of the PPU compared to the PCI-E bus PPU...but then again, you don't trust my words, so I see not point in continuing.

/out
 
Atech said:
I run quite a few things on the PCI-bus and that has never reulted in lower performance of the PPU compared to the PCI-E bus PPU...but then again, you don't trust my words, so I see not point in continuing.

/out
Click to expand...

:confused:

I was basing my comments on what I have heard over the years. Since PCI is a shared bus and PCIe point-to-point, it made sense to me. I didn't intend to attack you or anything...
 
One thing worth pointing out. When you're very near the limits of your Frames per second having a dedicated PPU is beneficial.

My friend has a Dell XPS M1730 with 8800M GTX SLI and an Ageia PPU. In UT3 when gaming at 1680x1050 with PhysX enabled on the GPU, 2xAA, 8xAF and all in game settings maxed it harnesses around 15-17FPS under the Lighthouse map.

Enable the Ageia PPU to handle PhysX and this increases to 25-35FPS.

Dedicated PhysX is the way to go for high res gaming where GPU bottlenecks might occur.

I'm in the midst of migrating to Windows 7 and have a 9800GT on the way to handle PhysX (replacing my Ageia PPU). ATi + nVIDIA works under Windows 7 and I intend on testing it.
 
Yeah, having a dedicated PPU will allow you to push things a bit further :) I see x16 PCIe-based GPUs as the most beneficial in this case, especially in the long run. It's also a great way to recycle that 'old' GPU you replaced ;)
 
I decided to try this dedicated gpu physx tonight. My main card is an eVGA GTX260 Core 216 SC (626/1350/2100) and the physx card is an XFX 8800GTS XXX 320mb (576/1512/1800). I just ran 2 passes of the Cryostasis physx timedemo @ 2048x1152 and @ 1920x1080 - once w/ gpu physx and once with dedicated gpu physx (8800gts):

2048:
gpu physx:

avg - 32.5
min - 19.8
max - 114.5

discrete gpu physx:

avg - 38.2
min - 22.6
max - 88.5

1920:
discrete gpu physx:

avg - 42.3
min - 24.0
max - 98.3

gpu physx:

avg - 34.9
min - 21.5
max - 117.3

So it seems to help. Of course this assumes that my two PCI-E 2.0 x16 slots didn't drop down to x8 / x8. The specs on the Asus maximus II formula say x8 / x8 with 2 gpus but I'm not sure if that's just in crossfire mode (it's not an sli board). Anyone know if both cards are staying at x16?
 
prasvt said:
So it seems to help. Of course this assumes that my two PCI-E 2.0 x16 slots didn't drop down to x8 / x8. The specs on the Asus maximus II formula say x8 / x8 with 2 gpus but I'm not sure if that's just in crossfire mode (it's not an sli board). Anyone know if both cards are staying at x16?
Click to expand...

No cards come close to saturating x8 PCI-E...so it's irrelevant
 
You must log in or register to reply here.
Back
Top