Obviously not. Both current and temperature are key exacerbating factors of electromigration. Clockspeed afaik, is not. But its a false dilemma, because a higher clockspeed will result in higher temperature and require higher voltage, regardless of what your testing shows (hint: there are several temperature sensors in the GPU, you are measuring only one, quite possibly GpuIO which is the memory controller. Try checking the shader sensor).

And its not like videocards that are not used for mining ever die  

Its really simple, any silicon chip thats used will eventually die. Its a given. The only question is how long it will take. It could be centuries, it could be days. But stressing these chips will reduce their lifespan, particularly when overvolting and overheating. If you are lucky, you may never notice, but it doesnt change the facts.

I don't recommend you use Deepbit.
However good/stable they are, they can't possibly make up for the 10% fee they charge the PPS-mode miners.
Eligius is a SMPPS pool, therefore you will get very steady payouts - every single share you send will be paid for.
While Deepbit pays you 0.00003597 BTC per share, current reward on Eligius is 0.00003997 BTC.
How about you underclock your GPUs by 10%? No? Didn't think so. Then why stick with high fee pools?

I'm currently mining at Bitminter and have only 0.13% rejected shares. That should put your 1.63% in perspective.
Using BitMinter, you have to get ready for some variance. There can be days when no block is found and you get nothing.
That shouldn't matter, however, as over time expected revenues from both Eligius and BitMinter converge on the same value.

To get back to your original question; running at high fan speeds is more likely to kill the fan prematurely. Running at low fan speeds (assuming that results in high temperatures) is more likely to cause premature failure of the GPU. Its impossible to put numbers on it, premature could mean 15 years instead of 150, it could mean 2 months instead of 20. But replacing a fan is fairly cheap and easy, replacing the GPU is less so.

Why?

From everything I have read, mid-80's is fine for GPU's......

I will have to lower the ambient temp in the rooms and see if I can get it down that way.  

cooler room and more space between the cards?

which cards?

have you heared of undervolting?

Dont think so. But there are other reasons to consider it as your miner app, like failover pools.

Я бы собирал на 6950 и куче райзеров, с несколькими недорогими БП. Ферма это растраты, не надо собирать гейминг-ПЦ в качестве фермы и уж тем более не с сильверстоунами и 1500ваттными БП.

Собирайте говнище старое, на старой материнке и селероне, ставьте туда видеокарты на железный каркас и пускай оно работает. У меня так собрано, работает уже с июня и ничего... Смысл в профите. А железо дешевеет.

Yes, I made on myself. First it will(depending on the miner, in my case it's cgminer) request via RPC to stop the offending thread. Should it fail, it will kill the miner altogether. It will then bench if the temp is still going above the threshold, in which case it will shutdown the PC.

It also detects FAN failure, and immediately shutdowns PC should it stop, or decrease below 700RPM.

I've posted this a few times actually, and before I realized it, I was bragging



Of course, it is for Windows only.

cgminer has built-in temperature and fan control.
I dont think it can shutdown the rig, but it will control fan speeds to maintain a desired temperature, decrease clocks automagically if temps exceed a given treshold and stop the GPU altogether if it exceeds another treshold.