Printable View
Got the 2 gtx 460 cards installed and running PG. Oc'd using Afterburner to 898 core, 1796 Shader, Mem at 1848. I did not touch the Core voltage and temps are at 40 and 50. One gpu keeps dropping from 898 to 405 then back to 898.
Where should my settings be? Do I have to increast voltage? Can I lower memory to help with heat? What is too hot with these cards? Thanks for the help.
makesure you are not running drivers 270/ revert back to 266. The 270 drivers throttle the GPU. Sound's like your issue. if you are already running the 266 drivers please post and I will look into other issues but this is one that is plauging the team right now.
This is the driver I'm using and it is working well for me
http://downloads.guru3d.com/Forcewar...load-2678.html
Yes. Unlock the voltage and up the voltage a few notches. They will run at 900 core / 1800 shader; drop the mem down to like 900 as well (no reason to have it sky high for the heat if you're just crunching on these), and you'll stay very cool.
898 1790 Voltage at 1000 temps at 54 with mem down to 920. Not sure how high I can safely. 2nd gpu driver keeps crahing at the above settings. Dropped it down to 850 and it's stable. How high could I up the voltage under Afterburner?Quote:
Originally Posted by DrPop
Not sure about Cuda projects.
What other Boinc project should I set as the backup?
When running PG, do I just run PPS Sieve or run Gcw Sieve as well? (Max credits)
Just a bump to a couple of questions I had.
How high could I up the voltage under Afterburner? Voltage set at 1000 mV
What other Boinc project should I set as the backup?
When running PG, do I just run PPS Sieve or run Gcw Sieve as well? (Max credits)
Just PPS Sieve for max credits. As far as a backup, I would actually say Collatz - reliable, decent CUDA credit, the backup setting works, and the WUs are short enough. Not sure about voltage.
Just a thought....are the two boards the same model #, same mfg, same bios?
Could explain the difference in OCing them.
Yo-
Also, there is sampling error. Some things do well, some things don't. Even things manufactured to the same specs don't all work the same. Welcome to the wonder and horror of statistics.