[snip]
About queue, expiry, scantime: Keep them default. I don't touch them and i have 0 issues with connectivity.
About worksize: Keep this @ 256. It has nothing to do with the pool difficulty, but rather GPU work unit. HD 7xxxx series cards appear optimized for -w 256.
About gpu threads: 7950s love gpu thread @ 1, with a high(er) Thread Concurrency number. 7950s appear optimized to run with 1 gpu thread and a high concurrency. Ensure that your thread_concurrency number is a multiple of 2 (2^n).
About vector: Unused, good for delete as you already have.
Your primary GPU related settings are: intensity, thread_concurrency, worksize, and gpu_threads. When you're using the config file format, tune those and leave everything else at default (unless a pressing need presents itself.)
I would also recommend setting your GPU fan to a constant 80-90%.
Thanks for the insight.
For now, I'm seeing low 70's for temps using the autofan to throttle the RPMs, which should help to preserve the fan life. I also have, left over from another project, three 5" Pabst 120 V.A.C. computer-style ball bearing fans running through a rheostat, positioned above the cards, feeding them ambient air which helps a lot to keep temps down. Once I ramp up the intensity, I figure that the gpu fans will ramp up anyway, but I'll see if I have to increase the max rpms or not at that point.
I got the thread concurrency number in my above config by following some instructions that I found on the internet and checking the .bin file name, and also off of the GPU comparison wiki, and used the one listed for my card. The number in my above config seems to be the max that I can use, or else the rig will crash.
I will leave the worksize and gpu thread numbers as they are, also. Thanks for the clarification.
The default queue, expiry, and scan-time numbers were 1, 120, and 60 respectively. If I leave those at default, won't it be bad to have work sitting in a queue, and also bad to have my cards checking shares for a long time that might become stale as they are being checked? Or is it bad to check them too quickly for some reason? I'm still trying to figure a lot of this stuff out so bear with me.
Another question: If I am looking at the Middlecoin work graph for my rig, I see some instances where the red balance line flatlines. Is this an instance where the server is down, or is this where my cards are doing work and not putting out valuable shares? I compared my chart against other peoples, and I can find identical flatlines, just trying to figure out what those mean. I'm pretty sure that someone explained this somewhere in the thread before, but a search didn't bring it up.