Ah ha! Now I know why random cards have dropped out. It's always happened when I've moved a fan or slightly adjusted the position of the M board, etc.
Yeah. I couldn't figure out why my rig did worse in the garage with a loud box fan blasting on it with near-freezing air temperatures (as opposed to inside the house with a less obnoxious fan setting). I finally narrowed it down to the boards wiggling around slightly. Sometimes I could get them in the right position and they'd run for hours and then randomly crap out. Here was my fix (just cut slits with a miter saw):
Look ma, no heatsinks:speed:13670 noncerate[GH/s]:
620.895 (2.425/chip) hashrate[GH/s]:629.412 good:43369 errors:455 spi-err:10 miso-err:0 duplicates:0 jobs:291 cores:78% good:254 bad:0 off:2 (best[GH/s]:626.034) Sun Nov 10 08:36:21 2013
board-2 speed nrate hrate good errors spi-err miso-er duplic good bad off per chip good cores
0: 863 38.426 39.309 2684 88 2 0 0 16 0 0 (2.402/chip) 77%
1: 862 39.399 39.456 2752 9 0 0 0 16 0 0 (2.462/chip) 80%
2: 809 34.388 36.444 2402 67 0 0 0 15 0 1 (2.149/chip) 71%
3: 863 38.898 39.752 2717 12 1 0 0 16 0 0 (2.431/chip) 80%
4: 859 39.313 39.900 2746 14 1 0 0 16 0 0 (2.457/chip) 79%
5: 863 41.905 40.482 2927 22 2 0 0 16 0 0 (2.619/chip) 81%
6: 861 38.640 40.831 2699 15 0 0 0 16 0 0 (2.415/chip) 80%
7: 862 40.487 39.932 2828 5 0 0 0 16 0 0 (2.530/chip) 80%
8: 863 39.786 39.425 2779 19 0 0 0 16 0 0 (2.487/chip) 79%
9: 859 40.659 39.805 2840 9 1 0 0 16 0 0 (2.541/chip) 80%
A: 863 40.630 40.101 2838 8 0 0 0 16 0 0 (2.539/chip) 80%
B: 861 39.256 39.594 2742 7 0 0 0 16 0 0 (2.454/chip) 80%
C: 855 38.469 39.161 2687 22 1 0 0 16 0 0 (2.404/chip) 79%
D: 809 34.775 37.438 2429 132 0 0 0 15 0 1 (2.173/chip) 69%
E: 858 37.209 38.717 2599 9 0 0 0 16 0 0 (2.326/chip) 79%
F: 860 38.655 39.065 2700 17 2 0 0 16 0 0 (2.416/chip) 78%
Keep in mind that this was manually tuned and I have access to unlimited amounts of very cold air.
That's beautiful, what type of tuning did you do to get the ones over 40 like that?
I went ahead and made myself a stablizer to attempt the same thing. I think because I had the 120mm fans on top sitting in a way to keep them from doing the shaking was why I was getting stable numbers.
Also, what is the measurement of the spacing you used on your board there? I think I will build the same thing. I used to foam which is working well but Id like to make it better.
