Does heat leakage increase with over-clocking?

[BonzaiDuck]

It's been a while since I posted here.

My friend across town built his system with a 3.2C and an ASUS P4C800-E "dee-lucks". My system consists of a 3.0C and ASUS P4P800. His uses Mushkin low-latency DDR433's -- 2 GB worth. Mine uses 1GB of OCZ EL Gold DDR500's. His system uses a Lian-Li midtower; mine uses a recycled Gateway (1995) full-tower.

I over-clocked my system to 3.6 Ghz, with the DDR500 running at 480 Mhz. Both systems have been operational for more than a year. The hardware is certainly not the latest.

My friend asked why I leave my system at the 3.6 setting. I explained that the over-clocking leaves all the voltages at their stock settings except for a 0.1V increase in the chipset/AGP voltage, and that my idle and load temperatures at 3.0 and 3.6 do not noticeably differ. My idle-to-load "delta" is 13.7C.

He says that his system heats up noticeably when he over-clocks it. We're trying to figure out what he may have overlooked, but his 3.2 should be "set it and forget it" at 3.8 Ghz. Well, qualify that with the caveat that his memory is only rated at DDR433.

But the motherboards are same generation. HIs P4C800 should only differ from mine with advantage. True, the chipsets are different, but nevertheless Intel chipsets and "same generation."

Four questions:

Question #1 -- What could be his difficulty temperature-wise with OC'ing his system? His system is so loaded up with fans that at full bore, it sounds like take-off for a C5A transport. Excluding my PSU, my system has twin intake, twin exhaust and a single CPU fan. I use a ducted motherboard, Zalman VGA cooler, with an XP120 CPU cooler running a Delta 120mm fan at just over 66% of its top speed. My friend uses a CNPS-7700-Cu -- a unit with thermal resistance greater than mine by only about 0.02 or so.

Question #2 -- possibly relevant to question #1: Does thermal design power -- heat leakage -- increase with an over-clocked setting even if the voltages are left at default?

Question #3 -- If my friend failed to take his "AGP/PCI ratio" setting off "Auto", could this contribute to his temperature variation from OC'ing?

Question #4 -- Could the temperature difference between stock and OC setting be explained by my friend's choice of memory, the memory's recommended VDIMM and the amount of memory? I think he said the Mushkin VDIMM he uses is closer to 2.9V while my OCZ's are set to 2.75V. I don't know his stock VDIMM setting, but my OCZ's are supposed to run at 2.75V as a stock setting, and I didn't have to change it.

 

[dguy6789]

Remember, they are both different motherboards. Most motherboards differ on what they report temps as. Remember, more fans generally are better, but the placement of the fans is more important than the amount. Make sure he is getting good airflow over his cpu, and there are the same amount of fans pushing air in as out. Remember to count the powersupply into the equation.

I am not aware of increased heat leakage at higher than stock settings with the same voltage.

 

[BonzaiDuck]

<< Most motherboards differ on what they report temps as. >>

Sure, but they should still be accurate enough to show the idle-to-load range with some accuracy.

I appreciate your observation about increased heat-leakage at over-clocked settings with stock voltages. How else could I conclude -- and validate through independent sensor -- that there is virtually no change in the idle and load values in stock versus clock settings? It has to be the case then that the leakage does not vary proportionately between stock and OC'd settings under the same load, even if the OC'ing increases power consumption. But with no confirmation from authoritative sources, all we have to go on is data comparisons of our various experiences.