By "thermal runaway", meaning non-linear temperature progression with increased heat-load. Rather annoying to have a system that is apparantly stable for most thing, then you go and run some program that loads up the CPU and causes it to generate 10W more heat than it generally does, and then the CPU ramps in temperature by 10C, and suddenly the CPU becomes unstable whereas on normal air-cooling or water the CPU would only heat up by an extra 2-3C or so and potentially remain stable.
Ls7corvete, TEC's have been in use by people for a long time. They are effective, but they do have their limitations. Yes, the newer lot of >250W rated TEC's makes cooling modern CPU's possible, but the caveats are still the same. Either you balance it correctly like you've done and remain within the narrow operational range, or you go all out and do the whole condensation protection thing, add a pump, feed the TEC more voltage causing it to suck down more power, boost the PSU providing that power, and so on.
Of course we read as far as the pump comment. What, you think that people who have been using TEC's for the last 6 years suddenly developed amnesia and forgot all the lessons they had learned until you came along?
TEC's can be effective, but they need to be properly controlled. Some kind of active temperature monitoring and power regulation feedback controller would go a long way towards making them more attractive. As you stated though, that now involves adding a pump, the power controller, and the extra power usage (which has to be paid for in the long term rather than up-front once). Could all of that be purchased for less than a high-end water setup, maybe, but there's nothing stopping someone from assembling a budget water-cooling setup with the same parts used in the TEC setup, and they didn't have to pay for the TEC and the ongoing extra power costs, and they don't have to remain within the TEC's effective power range.
So really it's a case of carefully balancing a setup, like you have done, and so long as you stay within it's limitations it's all good, or alternately going for the whole-hog TEC deal (pump, radiator, PSU, waterblock) just like has been done for at least 6-7 years now.
What I don't understand though is the quite blatant hostility you're showing towards other solutions, including those manufacturers who have been offering TEC-based water-cooling solutions for the last 5 years (i.e. Swiftech).
Your overclock (280x9) is rather middling. There are people with 3000+ Winnie's at 2600MHz on air-cooling, and 2800MHz on straight water-cooling, so a 2520MHz CPU clock isn't really demonstrating anything superior at this time, and is something which can be achieved by a near-silent heat-pipe based HSF. Winnie's are quite peculiar CPU's in that they don't overclock much better between water and air. A better indication would perhaps be a San Diego based core which does benefit from extra cooling. I would love to see some results from that, and more especially, a 3GHz CPU clock with the TEC being used, and what the power draw and total equipment was that was required to achieve that.
|