GTX 570 PPD discrepancy

C

craziplaya21

Weaksauce
Joined
Apr 10, 2010
Messages
81
I am really baffled here, according to what I'm seeing from other 570s, I should be getting much higher PPD than what is shown in my screenshot below. I am not even using my "gaming" OC (940/1900). Instead, I am using a more modest 885MHz. According to my OC and what I am seeing from other people's OC and their PPDs, I should be getting well over 17.5k, right? Drivers used are 263.09

weird_ppd_570.jpg
 
Last edited:
I have 2 in SLI in my sig rig, the primary gets 13-14.5 KPPD, the second gets around 17, 1625 shader clocks, same drivers as you. Disabling Aero does not help, running Firefox 3 vice Firefox 4 does not help. I'm really not sure what is going on with my GTX570s, but it might be related to your problem.

nice shader clock. my 17KPPD card, runs a couple C hotter than the other and crashes WUs if once in a while if I clock it higher than 1625. I just assumed that the hotter was because it is sandwiched between the other GTX570 and a GTX260.

the thread I started on my troubles is here.
 
Last edited:
An oddity that I did notice is that EVGA's E-LEET tool (or something like that) shows that Vcore for the GPUs are different, but I did not twiddle with it.

another note on my cards: matched pair, sequential serial number.
 
it's at 935/2000 1.1v 62c stock cooler.. going under water soon
 
really makes me wonder why I can only OC to 1625......

Maybe I should bum Vcore and push harder...

Really makes me wonder why our troublemaker cards are scoring so poorly compaired to other cards.
 
Something is definitely amiss there, I am pulling almost the same ppd on my 470 at 800mhz with the 260.99's. Do you have any video playback going on or flash sites loaded in your browser? Those lower my output considerably.
 
I don't think that is my issue, doing nothing with the rig, or cruising the internet, same results, with and without Aero, with and without WinAFC. Doesn't make sense.
 
jfb9301 said:
I don't think that is my issue, doing nothing with the rig, or cruising the internet, same results, with and without Aero, with and without WinAFC. Doesn't make sense.
Click to expand...

Did you try disabling sli, maybe there is some bug there.
 
i have a feeling its something driver related causing the issues.
 
weird..

i'm still on 263.09 if it matters

havent tried new drivers yet waiting for official release of 266.44
 
koven said:
weird..

i'm still on 263.09 if it matters

havent tried new drivers yet waiting for official release of 266.44
Click to expand...

Well, I also had the same problem with 263.09. I updated to 265.90 quadro drivers (with modified .inf) hoping it would fix it, but it didn't. Weird thing is, this issue only applies to F@H. The card is flying as it should when I put it through 3dmark 11, stone giant, metro 2033, and crysis benchmarks.
 
craziplaya21 said:
Well, I also had the same problem with 263.09. I updated to 265.90 quadro drivers (with modified .inf) hoping it would fix it, but it didn't. Weird thing is, this issue only applies to F@H. The card is flying as it should when I put it through 3dmark 11, stone giant, metro 2033, and crysis benchmarks.
Click to expand...


the problem is due to the card rendering something in the background. my guess is its something to do with aero and maybe some weird settings in the drivers making it even worse. i also noticed using custom desktop themes make it even worse. but ill take the hit over having to look at the default windows 7. also multi-display has an impact as well. gaming/benchmarks arent going to show anything that would even remotely compare to F@H performance.

though i cant really test anything to find the loop hole since i dont own a 500 series nor will i ever update past 258.96 drivers.
 
so here's something weird.. i picked up another 570 at fry's today

and this one is getting 16k PPD and it's a newer bios than my 19k PPD 570

the packaging they came in was different too

check this out

http://img153.imageshack.us/img153/4383/570q.jpg

they're both SC versions if it matters
 
Last edited:
the only difference i can see is the bios version

could that really cause it?

the 19k PPD one is on the old .72 bios
 
hmm what gives then? doesn't make any sense

mine are both SC, one is .72 , the newest one .74
 
and both of mine are the same bios....
Sequential Serial numbers, these are back to back identical cards.. and yet 2500 PPD difference.
 
just for shits and giggles one of you guys switch your display's to the second card and see if that one drops and the primary goes up. if the second card loses the same amount of PPD while the other one goes up then you know where the problem is.
 
ok, it seems a lot of people are having this problem


so here's the thing

i just ran my 16k ppd card by itself, it's still getting 16k ppd

i swapped in the 19k ppd card by itself, and sure enough, it's still getting 19k ppd

wtf! same drivers, same clocks

is it possible that some 570's just fold better than others? cus i dont know what else to do
 
only other thing is the bios screwing it up like yall suggested earlier. thats about the only thing i can think of. or your 16k PPD 570 really isnt a 570 and its not using all the shader cores. but thats a long shot chance.
 
koven said:
is it possible that some 570's just fold better than others? cus i dont know what else to do
Click to expand...
If they are identical and in perfect operating condition, no, unless when you swapped them they were installed in different PCI-E slots, which would be indicative of a motherboard related issue. Otherwise, only thing I can quickly think of would be the BIOS on the card with faster performance is highly optimized somehow. It's a strange discrepancy but seeing how I can't get two 8800s working at normal speeds in GPU3 even though they run the GPU2 client fine and fast, I expect anything is possible. /befuddled
 
it's so weird....... doesn't make any sense to me

i put them in the same slot when i tested 1 by 1

argh.. wish someone knew the problem/answer
 
Tomorrow when I shut my rig down to replace the 920 with a 970, I'll reverse my 570s. This should tell us if my problem is specific to one single card or if it is a PCIE slot/rendering issue.
 
it's definitely specific to the single card

i tried both by themselves..
 
At this point the only thing I can think of is to use nvflash to try the fast bios on the slow card.
 
I doubt it is the bios, the fact that I have the issue says the bios is not the cause
 
Would EVGA even accept a replacement RMA for this? I bought mine from Newegg and I can still go to them for a replacement, but I am wondering if it is better that I go with EVGA. Inside the box, it says not to contact the store I bought it from and to contact EVGA for any problems that may arise.
 
craziplaya21 said:
Would EVGA even accept a replacement RMA for this? I bought mine from Newegg and I can still go to them for a replacement, but I am wondering if it is better that I go with EVGA. Inside the box, it says not to contact the store I bought it from and to contact EVGA for any problems that may arise.
Click to expand...


go through EVGA and tell them you are having issues with your card on folding@home. dont say what team though. i'm sure they will do an RMA for it since they are heavily promoting F@H.
 
those that are getting ~16k ppd, what bios version are you using?

if someone here is on bios .72 and getting "low" PPD, then we can probably rule out bios
 
yeah.. my new card (16k ppd) is also .74


i know there's someone on another forum that has 16k ppd also, but with .73

hmm after exhausting testing options, im beginning to think it is the bios.. maybe something in the new bios causing it to get less ppd?
 
Dunno, all I know is for me to functionally identical cards at identical clocks on identical drivers get 2500 PPD different averages.
 
yep.. identical issue here

let me know if you find out anything..
 
2500 PPD loss would also mean your missing a huge chunk of shader cores. or cuda is failing to access all the shaders available to it.

the other thing that could be possible is that the gpu/shader clock ratio's are incorrect in some of the bios. so say the ratio's suppose to be 1:3 there might be a bios setting that is causing it to be 1:2.8 or something. just guessing but those are all possibilities.

(i honestly have no clue what the ratio is on those cards so i pulled random numbers out of my ass so dont try to correct me)
 
You must log in or register to reply here.
Back
Top