[insanedoc]

According to the bios and numerous monitoring programs, my slightly overclocked (20%) Intel E8400 idles at 60C and spikes to 70C under 80%-90% load. It has operated as such for the past several months entirely stable, but I'd like to push a bit further, to around 4 ghz if possible. However, I doubt it would be stable at higher voltage in these temps.

I'm using a copper aftermarket Zalman heatsink with arctic silver compound, and there are five 90mm case fans all pumping air out of my case (due to certain obstructions I can't really get good front to back airflow so I don't bother). Room is air conditioned.

I've tried refitting the heatsink several times, cleaning both surfaces with 91% alcohol, cleaning the dust out of the fins of the heat sink with compressed air, spreading the compound with some plastic, and not spreading the compound.

I've seen idle temps 40C and lower on similar air cooled set ups and clock speeds, yet mine is off the charts in comparison. Any ideas on what could be keeping my temps so high? thanks.

[I.M.O.G.]

What does Coretemp report your temps as?

Bad mounting could shoot your temps up that high.

How are you applying the thermal paste? What zalman heatsink exactly are you using?

Voltage?

If you have 5 fans pumping air out, where is air coming in from?

[burebista]

Originally Posted by insanedoc

According to the bios and numerous monitoring programs, my slightly overclocked (20%) Intel E8400 idles at 60C and spikes to 70C under 80%-90% load.

Did you run a prime95 Small FFTs for 15 minutes?
Take RealTemp and do a CPU Cool Down Test then post a screenshot here. Stuck sensors are very common on 45nm CPU's.
70C in load is perfect fine. What Vcore?

[insanedoc]

Originally Posted by I.M.O.G.

What does Coretemp report your temps as?

Bad mounting could shoot your temps up that high.

How are you applying the thermal paste? What zalman heatsink exactly are you using?

Voltage?

If you have 5 fans pumping air out, where is air coming in from?

Coretemp reports 60C idle as well.

here's the heatsink http://www.newegg.com/Product/Produc...82E16835118223

Its mounted with a back plate, there are 2 hooks on both sides of the bracket and a separate metal piece that presses the heatsink itself down, and creates pressure with the hooks. I dont know if that makes sense but you can probably see it in the newegg pictures.

I've tried putting the paste on a couple of ways, last time I just cleaned it off really well and put a rice grain sized amount of paste, and let it spread itself. I've also tried spreading it just a little with the tube itself so it's more even.

voltage is stock. I could set the clock back to 3.0 for testing purposes if that helps.

There isn't an intake fan, but the case itself is mostly mesh so air can easily get in. Thing is I've tried taking of the cover and putting a box fan next to it which didn't make any noticeable difference, I think it's a heatsink problem.

[insanedoc]

Originally Posted by burebista

Did you run a prime95 Small FFTs for 15 minutes?
Take RealTemp and do a CPU Cool Down Test then post a screenshot here. Stuck sensors are very common on 45nm CPU's.
70C in load is perfect fine. What Vcore?

It likely won't complete a prime test because it pushes the load to 100% and even higher temps, but I can try it out.
edit: Yeah it's hitting 82C in small FFTs test within seconds, I'm not comfortable leaving it on any longer.

I don't know if I have a different Realtemp but this one only has a Sensor Test and an XS Bench, is it one of those?

I realize 60-70C is stable but I want to OC it more in the future, and before I can safely do that I should probably get the temps down.

[I.M.O.G.]

Thanks for all the answers to my questions...

EDIT: just read your second response to burebista... Do the sensor test, its explained here:
http://www.techpowerup.com/realtemp/docs.php

That said, it still sounds like bad contact with the HSF seeing as how quickly your load temps skyrocket - is the heatsink butting up against any capacitors or anything close to the socket?

Are you sure you are using the socket 775 clips and not those for AMD? I don't think its possible you could be using the wrong thing, but if the clips are bent or damaged, they may not be appling enough pressure for good contact.

[EarthDog]

That cooler is more form than function (made to be quiet not to be the best on air). With that said, the temps are still pretty high as others have said.

When your temps are up that high, does your heatsink at least get warm to the touch? If not, then it is a contact issue.

If you push down firmly on the heatsink do the temps drop?

[insanedoc]

Originally Posted by EarthDog

That cooler is more form than function (made to be quiet not to be the best on air). With that said, the temps are still pretty high as others have said.

When your temps are up that high, does your heatsink at least get warm to the touch? If not, then it is a contact issue.

If you push down firmly on the heatsink do the temps drop?

To answer the above, I can't really run the sensor test because it stresses the CPU to uncomfortable temps. Also, the heatsink does appear to fit snug with plenty of pressure and is not touching any other parts.

Pushing down on the heatsink doesn't lower the temps any and although it's somewhat warm to the touch, it is not at all hot.

I discovered something interesting during a time when I tried spreading the thermal paste with some plastic really thin on the processor, before once again mounting the cooler. My temps skyrocketed to 80C+ under load, and after quickly taking it apart I noticed that the paste had only touched the very edges of the heatsinks base. Most of the area of the base was dry and still shiny as if I had just cleaned it. I also noticed that the corners of the heatsink are scratched, where as most of the surface is a mirror-like shine.

Is it possible that either the CPU or heatsink surface is so uneven such that they don't even make contact around the center?

I do have the stock heatsink I could try out for the first time to compare, but that would be a lot of work.

[I.M.O.G.]

That sounds like a contact issue of some sort... Some people lap to get their base even - if your base is convex and only touching the outer parts of the core lapping the heatsink base may be a good idea for you.

[insanedoc]

Originally Posted by I.M.O.G.

That sounds like a contact issue of some sort... Some people lap to get their base even - if your base is convex and only touching the outer parts of the core lapping the heatsink base may be a good idea for you.

Could the CPU itself be uneven?

[EarthDog]

Right. ^^

Try placing a razor across the cpu and block and see which ones arent true.

[I.M.O.G.]

I'd consider that much more rare than an uneven base, but ya, earthdog's suggestion is an easy way to figure out what is what.

[EarthDog]

Old C2D's (think e6600/e6x50) were notoriously concave IIRC...I also believe thats why some blocks/heatsinks were made convex, though that is just speculation on my part.

[insanedoc]

Originally Posted by EarthDog

Right. ^^

Try placing a razor across the cpu and block and see which ones arent true.

I'll try that the next time I take it apart, probably tomorrow. Thanks for the tips so far, I'll update asap.

[insanedoc]

Lapped both surfaces to little effect. Although I couldn't find sand paper any finer than 800 grit, so it didn't turn out super shiny. Not sure what more I can do.