GeForce 320.49 Slow?

[Kurtis200200]

I'm actually running at 90k PPD on the 320.49 build, using two GTX 660s on client-type advanced, so no complaints about that driver set at least for F@H, though admittedly the driver could still be bad for gaming, I wouldn't know

[codyf86]

You're getting 90K PPD on advanced with two 660's? Is the 90K PPD on the 8900 WU's? Two 660's should be getting well over 100K PPD doing 8900's. One 670 gets around 80-95K PPD by itself and the 660 shouldn't be getting much less...maybe 60-80K PPD per card.

I've been getting around 90K PPD with one 670 on the 8900's. The difference between the two cards is noticeable, but nothing astronomic. Seems like something isn't right.

[Kurtis200200]

90k PPD total on P8900 WUs using two evga GTX 660s (02G-P4-2662-KR), yes.

According to the GeForce 600 Series page, the 670 has 2460 GFLOPs, while the 660 has 1881.6 GFLOPs. While I'm aware that may not be the correct metric, if the proportionality holds true for other, F@H-relevant metrics, then the 660 should be getting 1.4x less PPD than the 670, or ~64k each.

I have a Q6600, and also run BOINC but leave two cores alone to feed the graphics cards.

I also have 4gb ram, though I'm frequently on resource monitor and even when running two P8900s and two Clean Energy Project phase 2 (BOINC) work units I have free memory (not just available, free).

The graphics cards on P8900 average around 72°C, maybe getting up to 74°C peak.

I think the hard drive is a 7200rpm.

[codyf86]

Ah yeah,

RAM doesn't really matter as FaH GPU WU's use almost no RAM, but it's more-than-likely because your running BOINC also. No biggie 90K PPD is still pretty good for running FaH and BOINC together.

[Kurtis200200]

If I could shove two more 660s on my motherboard I would.

Also, according to http://compdewddevelopment.com/projects ... /index.php
(and depending on the project), for Core 15 22.5k PPD appears to be the benchmark for gtx 660, and my gtx 660s are now working on P8054 (unfortunately no data there) and pulling an estimated 22k PPD each

[codyf86]

Yeah,

My example above was only for the 8900's; not really sure about other WU's. Seems like your getting good numbers now.

[Napoleon]

According to the GeForce 600 Series page, the 670 has 2460 GFLOPs, while the 660 has 1881.6 GFLOPs. While I'm aware that may not be the correct metric, if the proportionality holds true for other, F@H-relevant metrics, then the 660 should be getting 1.4x less PPD than the 670, or ~64k each.

Actually, PPD is not the correct metric to apply 2460/1881.6 (or 1881.6/2460) to because it is nonlinear due to QRB. Applying it to TPF would work better, and then you could calculate PPD difference based on TPF difference.

[rbpeake]

Hello all,

Just wanted to share my observation. The 320.xx series Nvidia drivers seem to be problematic for a good percentage of the people using them...for gaming and folding. I upgraded to the 326.01 drivers for Windows 8.1 Preview (They install and work fine on Windows 7), and they have been leaps and bounds better as compared to the 320.xx series.

My 320.49 driver fails after a day or two of use, and then it would be some time for the driver to recover (if at all).

I installed the beta 326.19 a little while ago, and am keeping my fingers crossed that it works better (by keeping working!).

I just replaced my PSU about the same time the driver began failing, so I was worried the problem was with my new PSU!