Quote from: skrazy on September 22, 2013, 10:00:46 PM

Quote from: klondike_bar on September 22, 2013, 09:32:37 PM

Quote from: dacman61 on September 22, 2013, 08:52:32 PM

Quote from: greaterninja on September 21, 2013, 04:23:15 AM

Kit #1: 3 cards:
1: 30.609GH/s
9: 31.339GH/s
13: 29.335GH/s

Kit#2: 2 cards:
Bank 1
1: 30.723GH/s
9: 30.093GH/s

i penciled the resistor to about 1230ohms.

Should I have done that or go lower?

I was able to do 1.18k Ohm... I've been steady at around 33Gh/s.

I tuned down to 1.165k, getting around 35Ghash, and 33Ghash/noncerate at the pool. Amazingly, chip #13 which was running at ~300mhash with tons of errors now works just fine.

I had a similar issue with one chip on an h-card. When I changed the resistance to 1.2k the chip went from ~0.100 to 2.000. That said I did need to clock it lower than the rest of the chips to achieve a reasonable error rate. At 53 it throws about 4% errors.

hmm sounds like the exact same issue. i have that chip tuned at 53 now with 6% errors

However, i now find that after some operation 1 chip (usually random, but 6 or 7 seem slightly more common) will drop to 0 hashrate until i stop and start the webgui. What could cause this? I am wondering if the chip is being shut off due to the increased voltage. With all 16 chips i see ~37Ghash and 34Ghash noncerate. with 15 its 35Ghash and 32.5ghash noncerate

temperatures seem okay. they are only slightly warmer then the original was before heatsinks and modding, and have significant airflow. (i read that fan/airflow vibrations can cause issues - is this actually a thing or an annecdote?)