These 1888pt GPU WU's are killers. I had to change some things around last night. One of my K9A2 MB's was a bit doggy to begin with, and with 3x cards flaming away on 1888's it kept amping out and finally hit the skids last night. I was wondering why I couldnt' get nvclock to monitor the cards on that board properly, now I know why.
I think I've got everything ticking along ok now. Still need to re-image my file / media server, will pribably do that later this evening. Adding 4TB of storage and moving the LVM around a bit.
Now I've got 2x Q6600's and 1x A9850 that I need to find system's for. May just sell them, as keeping up with 4 boxes is enough work

.
I use the same flags on all the GPU Boxes and set -advmethods while doing -configonly and -Big WU's:
wine
[email protected] -gpu 0 -verbosity 9 -forcegpu nvidia_g80
SMP, same thing for -advmethods done duing -configonly w/Big WU's but only two flags when stating the WU: (-smp and -verbosity 9 )
EDIT:
-forceasm
Instruct the core to use SSE/3DNow assembly instructions if possible, even if it has previously made the determination that the machine may not be handling this well.
I'm not sure this has any real meaning anymore but I do see "SSE Extra-Boost" being flagged out on some WU's but not on the A3's thus far.
.