Seti@Home optimized science apps and information
Optimized Seti@Home apps => Windows => GPU crunching => Topic started by: M_M on 20 Jul 2011, 01:55:28 pm
-
I have noticed a problem with GPU (GTX560Ti, well cooled, CPU temp never ever above 65C) that sometimes, during x38g execution (happened sometimes also with older 0.37 app), it donwclocks to low-power state 405MHZ GPU, 810MHz Memory, and will not upswitch back until machine is rebooted.
Doesn't happen during the gaming, but happen from time to time during seti crunching. Drivers latest official and stable 275.33.
Is this some known fault/incompatibility with lunatic application? Does it happen to anyone else? Is there some workaround?
-
I have noticed a problem with GPU (GTX560Ti, well cooled, CPU temp never ever above 65C) that sometimes, during x38g execution (happened sometimes also with older 0.37 app), it donwclocks to low-power state 405MHZ GPU, 810MHz Memory, and will not upswitch back until machine is rebooted.
Doesn't happen during the gaming, but happen from time to time during seti crunching. Drivers latest official and stable 275.33.
Is this some known fault/incompatibility with lunatic application? Does it happen to anyone else? Is there some workaround?
It looks like an issue with the 560ti model, there'sThis thread (http://setiathome.berkeley.edu/forum_thread.php?id=64807) over @ Seti where Slavac is\was asking for information
As far as I know there may be some issue where the card is being pushed to hard past reference Nvidia spec's
There's also an issue with the Boinc API where if the application (Boinc) doesn't exit a task nicely, it will downclock itself to protect the card. This "feature" is only apparent in Cuda 4.0+ drivers. Cuda 3.2 drivers (pre 275.xx) don't exhibit this behaviour