Folding Forum

davidcoton wrote:For QRB you need to get a passkey and then return at least 10 WUs with 70% (from memory) success rate.
It is still helpful to post your log (including the system and config sections), so people here can see if there are any other reasons for low ppd.
I agree with billford about what a 780Ti can get

Edit by Mod: Correction. Your total successful returns of bonus-eligible WUs must be >80%

I didn't notice that the QRB only works with a passkey - thanks!

But I'm dealing with a new problem now: Almost every GPU WU is returning with an UNSTABLE_SYSTEM error. Is F@H sensitive to overclocking (I have a factory OC Card with 1215Mhz boost frequency)? Furmark runs without any problems so far. Does anyone know what to do, maybe reduce boost clock?

PPDGuy

Yes, also factory OC can cause trouble. Reduce OC is a good idea.

Also, it may be running too hot- on mine the default fan profile wasn't good enough for folding. Increasing the fan speed (and limiting the maximum allowed temperature) means that I can now run it with another 25MHz on top of the factory overclock.

Yes, the error checking in the folding core will detect any minor calculation error that overclocking or overheating causes and fail with an unstable system error. Folding uses different elements of a GPU than Furmark, so Furmark is not a good indication of stability. A better test is FAHbench as it uses actual folding code.

Also, some have reported that lowering the memory clock on a GPU can leave a bit more room for upping the GPU clock itself.

billford wrote:Also, it may be running too hot- on mine the default fan profile wasn't good enough for folding. Increasing the fan speed (and limiting the maximum allowed temperature) means that I can now run it with another 25MHz on top of the factory overclock.

I'll second that suggestion. My 750 Ti (no overclock) had been rock-solid for quite some time and suddendly it started dumping WUs consistently. It started on a very hot day and I suspended it. Then continued after the weather was cooler. Not knowing if it had failed or something else had happened, I moved it to another machine. (It had been in the next slot to a 650 Ti). The problems stopped -- so now I need to put it back and work on the fan profile.

bruce wrote:so now I need to put it back and work on the fan profile.

I found it helps to pay some attention to the airflow patterns within the case too. I added an extra fan behind the intake grille to blow cool air into the case (instead of relying solely on the extract fan to pull the warm air out), that gave me a few degrees.

I was running the case with the side off and discovered, completely by accident, that if I closed off the top two thirds (approx) of that side then the gpu runs 2 or 3ºC cooler (it's a tower case with the gpu horizontal near the bottom). Haven't really worked out why, but I'm not fussy about theory if the practice works. It was stable anyway, but those few degrees meant I could overclock it a bit more

Obviously other setups will be different, but a bit of experimentation can give a useful reward in terms of overclock and hence PPD, for no cash outlay!

(I don't count the cost of the extra fan, that was trivial)

Another thing to look at is airflow provided by the fans. I found over the past couple builds that stock fans in some (Most? Even Lian Li!) cases are anemic compared with the Noctua fans I replaced them with. I am a fan (pun intended) of "positive pressure" rigs where the intake fans have higher capacity than the exhaust fans. One reason is that it reduces the amount of dust sucked in from nonfiltered crevices in the case. It is also easier to experiment with extra exhaust slots such as vented backplane covers, to direct the flow where you want.

I downclocked my GPU and the UNSTABLE_SYSTEM Errors are gone. But now i get BAD_WORK_UNIT errors quite frequently.

******************************* Date: 2014-08-06 *******************************
03:47:29:WU01:FS00:0x17:ERROR:exception: Error downloading array velm: clEnqueueReadBuffer (-5)
03:47:32:WARNING:WU01:FS00:FahCore returned: BAD_WORK_UNIT (114 = 0x72)

I also noticed that after that error my GPU is stuck at 549Mhz. A system reset is the only solution to go back to normal clocks.

When your machine forces a very low clock rate that can only be solved by rebooting it is invariably a sign that your driver did a GPU reset (which also tends to kill the WU that it is working on) which means the GPU quit responding to driver requests. Resets show temporarily in the systray but disappear. However, they are also permanently recorded in the Windows system event log and you can check for that. It all mean that your card is not yet stable enough for folding. Solutions include: even more down-clocking; more cooling if your GPU is running hot; more voltage; Don't game while folding.