DangerDen 6800 water block

[bellevegasj]

pretty similar results from what i have kungfooj.

i'm wondering if it's even worth it to water cool the gpu. i've seen lots of people's over-clock numbers using water-cooling that aren't much higher than ours.

[msv]

Heat removal ==> case temp, power circuitry temp, MB RAM temp, hdd temp...
Basic question, basic answer. Sorry for the grumpy tone.
Besides, could the graphic RAM handle 1185 MHz under *any* circumstances? Very small peltiers?
regards
Mikael S.

[aaronspink]

Originally Posted by kungfoojesus

Here's what he wrote to me about killing the 6800GT when I emailed him.

To keep it short, I used the nvidia "determine optimal clockrates" utility to initially overclock the card. (mistake number two) It set a rate of 485/1185, and one run of AquaMark later, the card was history. It started artifacting about halfway through the bench, I immediately killed AM and reset the clockrates to default, but the damage was already done. It really can happen that fast: less than 30 seconds elapsed between the appearance of the artifacts and my successfully resetting the clockrates."

Take it for what you will but this guy has been water cooled since 97 and overclocking so I'm defnitely going to take his advice and put a case fan over those ramsinks when I put on my silverprop. I definitely won't be putting that 3 pound dangerden monster on a video card though. Even if it does WC the ram.

Listen, I know semiconductor electronics and you aren't going to "fry" your card by changing the clock frequency of either the memory or the gpu. Not going to happen. Not physically possible. At best, uping the clock rate on a semiconductor device will result in incorrect results, but the results will be correct when the frequencies are set back to the factory tested defaults. The only major post manufacturing issue in the actual silicon is electromigration and EM is influenced mostly by voltage with a secondary of temperature.

Two, the dram chips used on something like a 68GT don't produce enough heat and have a high enough package thermal resistance that the difference between fully passive and water cooled will be minimal at best.


Aaron Spink
speaking for myself inc.