Sign in with
Sign up | Sign in
Your question

Hyper 12 plus cpu I7 2600k temperatures

Last response: in CPUs
a b à CPUs
April 22, 2011 10:31:48 AM

I get the following temperatures in full load 4.2GHz 65C
5GHz 95 C whit the minimal voltage where is start to work. I suspect something wrong whit hyper 12 plus maybe termal contact.
a b à CPUs
April 22, 2011 10:47:02 AM

Those numbers look about right to me. It's a good cooler (assuming Hyper 12 plus = Hyper 212 plus), but it looks like you might be expecting a bit much.
a b à CPUs
April 22, 2011 12:17:12 PM

As frequency increases, so does power draw and consequently the amount of heat generated. If you graph out the power drawn and temperature increases as you ramp up frequency in 100mhz increments you'll find it is not nice and linear. There will come a point where power & temperature are increasing VERY rapidly. When you get there, that's pretty much the limit for your cooling system.....

Ah, that said, if you into OC'ing to extremes you'll need to realize a hyper 212 isn't the right cooler for the job. Do a bit of research and pick out the best air to air you can buy. Maybe something like this for starters:

Or go to water cooling...

That will lower your temps a bit but I'd also research a bit on max OC, I've seen 5.8Ghz on liquid nitrogen.... (but there's always someone faster)
Related resources
April 22, 2011 12:18:11 PM

if you really want low temps just get a 3x120mm H2O rad and a nice cpu block and those temps will not go over 480sli and i7 2.66@3.8 never go over 55 under full load
a b à CPUs
April 22, 2011 2:32:38 PM

Yeah, for high-clock high-voltage situations the Hyper 212 Plus isn't the right cooler. It's good, but not that good. Get one of these:

Noctua NH-D14
Thermalright Silver Arrow
Prolimatech Megahalems Rev.B

Coolers like those can handle the heat generated at full load when doing the over-volting necessary to get 5GHz or higher overclocks. Or go with a nice water cooling loop, like an XSPC Rasa kit.

Also note that you should stay below 1.4v for CPU Voltage and below 75ºC for temps on an everyday overclock.
a b à CPUs
April 24, 2011 1:22:23 AM

Thanks guys. For 5 GHz over clock I need 1.445V and at 4.2GHz gist 1.325V on I7 2600k. Some approximation formulas as I know:
P = (V^2)/R , F= V, P = F(V^2); results P= (O)^3.This is from theory. where "=" has meaning of proportional in the last tree formulas.
But the maximum overclock frequency is not flowing the direct proportionality whit the voltage. Is like something exponential. May be the PLL oscillator douse not work well and is not stable enough at the hight frequency around 5 GHz.The Hyper 212 Plus has a not very smut surface at the contact whit CPU.Can this cause overheating?
May be some Mather-boards are better and anther . Therefor is someone knows some tricks whit the settings and will like to share. Or may be the power vcore rail is not god regulated at the high load. I relay do not know way the maximum overclock frequency douse not go linear whit the voltage.
a b à CPUs
April 24, 2011 3:26:31 AM

That's actually normal -- it won't scale like any formula says. You'll reach a certain speed and it will require a lot more voltage to be stable. Usually, that means you're either at or near the top overclock speed the CPU will give.
a b à CPUs
April 24, 2011 12:45:55 PM

Ah yes, EXPONENTIAL (logarithmic too...) changes in power draw ..... I did point that out in my last post.

If you want to be real scientific about it, follow the steps listed and plot out temp/voltage/power as frequency is increased. When you get on the the exponentially rising part of the curve, your Not going to be stable.

Yes, some motherboards Are better than others at overclocking. One main contributor to the better performance is more stable voltage to the CPU:
- either thru better voltage regulation (better inductors, caps, ...)
- or, thru nice heavy copper traces on the MB

Yes, some coolers are better than others
- larger surface area
- better fans

A better cooler or better motherboard Can move the exponential part of that curve out a bit further.... but, as always, going faster always cost more money