My baseline stock temperatures at 75F room-ambient are lower than any I've seen on this forum since we began this and other related discussions. I have ducted my motherboard, VGA cooler and CPU cooler in a pressurized case to twin Sanyo Denki San Ace 0.52A fans, so don't feel bad.
I want to add some comments here, of a sort of editorial nature.
To capture an additional 100 Mhz of processor speed, I make an educated guess that you've raised your VCORE about twice as far to get from 3.0 to 3.1 in comparison to the stable VCORE settings going from 2.9 to 3.0. Your load temperatures -- I assume your room-ambient is 80F or less (and I sweat profusely at 76F) -- range through -- and into -- the high-60s. Your VCORE at 3.1 is about 0.04V higher than mine at 3.0, but my VCORE at 3.0 is only 0.01V higher than my VCORE at 2.9 Ghz.
Another of our members recently blew his 680i motherboard, and now claims (I paraphrase for dramatic effect) that the 680i chipset is "for s***."
On Sunday, I met with my 1960's college-days dorm-hall brothers for a reunion in an LA suburb. Two of them are plasma-physicists. We casually discussed -- mostly for my benefit -- the issues.
"Heat dissipated increases by the square of the voltage divided by the square(?) of the resistance."
Ceteris paribus, electrical circuits conduct electricity better at lower temperatures. Thus, phase-change cooling and liquid nitrogen, refrigeration, evaporative coolers in water-loops, and high-wattage TEC-chillers. This also implies that lower temperatures allow greater electrical efficiency and higher clock-speeds with the same amount of power (and voltage) or less.
In consideration thereof, but with independent implications, both heat and voltage cause CPUs to degrade or fail over time. Voltage creates heat, but voltage by itself can cause failure and degradation.
The implications of the parabolic relationship between voltage and resistance, resistance and heat, heat and efficiency, suggest to me that we should limit our over-clocking vcore settings to a region with an upper-bound determined just at a point where a disproportionate increase in voltage is required to achieve an equal increase in speed, with due consideration to the point where voltage increases suddenly promote disproportionate increases in temperature.
Water-cooling, or any method which can bring load temperatures within a sliver of room-ambients will significantly allow the use of lower voltages to make the same level of over-clocks, and within the range of voltage increases that I advocate here, slight increases in voltage will yield higher over-clock settings that are stable.
That being said, you cannot water-cool to the limit of room-ambient while overwhelming the circuitry with voltages that go way beyond the intended factory voltage spec. And to blame the chipset or motherboard manufacturer just seems to exhibit a state of denial. As much as we may have budgets for frequent processor and motherboard replacements, I personally do not believe it makes sense to carelessly shorten the longevity of hardware to a few months, when it was intended to last several years at stock settings, and may last a majority of those years with carefully chosen over-clock settings.
This also holds for air-cooling. So again, you can cool down your processor to a point where a lower voltage supports the same clock setting and stability. You might then be able to raise the voltage again to its previous level and get higher over-clocks. But just because you can raise it even more to achieve "extreme" over-clocks while keeping the temperature from rising too much, it is possible to overwhelm the circuitry with too much voltage even as you may have lowered resistance through cooler temperatures.
Facebook
Twitter