A few questions about under-volting.

TechnoCorruption · 18 Jan 2008 at 18:58

I have been testing my E2160 (L2 revision/1.325 vid) and Asus P5K-E Wifi over the last few days and managed 3GHz with the stock cooler and stock volts. I have now bought a AC7 Pro and now manage 3.2GHz with 1.4v, I tried for 3.3-3.4GHz but it becomes unstable regardless of extra voltage.

So to cure my boredom I thought I would try some under-volting. The lowest my motherboard goes is 1.1v so I selected that and managed to boot at 400x6 for 2.4GHz. My temps have dropped drastically only reaching 39c (AC7 on ultra quiet mode - 800rpm) full load under prime. I fired up some Source based games which are known to be cpu dependant and I notice no difference at all.

Anyway here comes the questions... seeing as I have been getting load temps of 60+ during the overclocking and over-volting stages do you think I can use the AC7 passive? The fan is only spinning at 800rpm at the moment so surely it can't raise by more than 20c, can it? Also I was wondering if anyone had tried some under-volting in the past and what sort of results they got?

What started off as a bit of boredom has really stoked my curiosity, I am just not sure if when the summer comes it will be logical to be running on minimal fans with passive cooled processor and graphics card.

Oh and by the way... I am even considering updating my bios to the latest 0906 just to get the renowned 0.1 vdrop as I can't select any lower than 1.1v

Meridian · 18 Jan 2008 at 19:35

Simple answer: try it. Generally undervolting works well at default speeds, or a little above, and people do it for precisely the reason you give: to keep temps down. But once you get above the speed where you would otherwise need to raise the voltage to keep the system stable, then undervolting is likely to fail. It comes down to which is more important: speed or temps?

M

TechnoCorruption · 18 Jan 2008 at 19:56

Well it failed prime at 2.4GHz 1.1v so I upped it to 1.2v. It seemed to be happy in prime (around 45c load with the AC7 at 800rpm) so after 40 minutes I pulled the fan plug out.

It slowly hovered up to around the 70c mark before I plugged it back in. I know that day to day running of the computer will never stress the processor as much as prime does but there again its winter now... and the summer is another story.

Just for curiosity I will try the processor at its stock speed and 1.1v and see what happens. The difference between 2.4 and 3.2 could be dismissed in the search for a passive computer but I am not sure if I can dismiss 3.2 for 1.8

Meridian · 18 Jan 2008 at 22:06

I rather suspect passive is probably impossible with that CPU and that cooler - passive coolers for CPUs (like the old Zalmans) seem to have disappeared. Will be interested to know if it's possible on the 45nm CPUs though.

M

TechnoCorruption · 19 Jan 2008 at 09:27

Its definitely possible with these 65nm chips as long as you are prepared to run at stock. With the clock speed at 1.8GHz and the voltage at 1.06v under prime load I was getting temps of no more than 60c running the AC7 passive.

The only thing I did notice is that the temps take a while to drop when going from load to idle as there is no air to force cool the heatsink, I doubt it would matter much though as the only real thing to keep an eye would be maximum temperature.

As you say though the 45nm processors would be great to be run like this. If the new Wolfdale's do 3GHz at stock which is about 1.2v for 45nm then they are born for some under-volting. With the right heatsink they would be great for an absolutely silent HTPC.

Anyway my curiosity is satisfied now so its back to trying to persuade this chip to go more than 3.2GHz