Thermal paste

[cboath]

I've seen people commenting both ways.

What way is the best/preferred method for applying the paste? Pea size dab in the middle of the chip? Or a line across all 4 cores?

 

[daw123]

I use the Artic Silver 5 application:
http://www.arcticsilver.com/in...ute_step2intelas5.html

A line across the four cores.

 

[konigswagger]

line across four cores. a pea size in the middle for a quad doesn't make sense since the dies are not placed smack dab in the center

 

[TitusTroy]

the Arctic Cooling MX-2 instructions don't differentiate between CPU's...they only list 1 set of instructions---pea sized drop in the middle of the CPU...the Arctic Silver instructions seem more detailed based on the CPU (dual, quad etc)

I've always done the drop in the middle method but I'm willing to change it...is the line across the middle the hands down best way to do it for a quad core CPU?

 

[BonzaiDuck]

I'm still sticking to my decision for terminating use of AS5 in favor of IC Diamond.

Since I've made too many sales pitches here over the months based on my load-temperature sampling -- every 8 seconds for an hour at a time -- comparing IC Diamond with AS5, I'll skip that here.

IC Diamond is thicker and grittier than AS5. But for either product -- or any TIM for that matter -- I use a dollop that is just a tad larger than long-grain rice. I drop it in the center of the IHS and use a rectangular razor-blade to spread it over the IHS so that I can only see a layer of TIM with no metal showing through.

Since I lap the nickel or zinc off my IHS processor-caps, leaving them very flat, this method seems to work well for me.

 

[ZimZum]

Like I'm making a peanut butter sammich. Only spread it out very thinly.

 

[ZimZum]

Originally posted by: BonzaiDuck

Since I lap the nickel or zinc off my IHS processor-caps, leaving them very flat, this method seems to work well for me.

I never had the balls to lap a processor. Is there enough of an effect on temps to justify the risk?

 

[Rudy Toody]

I put a pea-sized blob of IC Diamond in the center of the chip, mount the heat-sink. and give it a ten second buzz with my beard trimmer.

When you buzz with the vibrations perpendicular to the plane of the chip, you will get a very nice circular spread with possibly a little pump-out around the edges. The IC Diamond is not electrically conductive so this doesn't hurt anything.

If you try this technique without fastening the heat-sink down and then remove it after the buzz, I think you will be surprised by how effective this is.

It sure beats using a razor blade, putty knife, or credit card.

 

[Rubycon]

Originally posted by: Rudy Toody
I put a pea-sized blob of IC Diamond in the center of the chip, mount the heat-sink. and give it a ten second buzz with my beard trimmer.

When you buzz with the vibrations perpendicular to the plane of the chip, you will get a very nice circular spread with possibly a little pump-out around the edges. The IC Diamond is not electrically conductive so this doesn't hurt anything.

If you try this technique without fastening the heat-sink down and then remove it after the buzz, I think you will be surprised by how effective this is.

It sure beats using a razor blade, putty knife, or credit card.

How does that work (specifically with a beard trimmer?) Would an electric toothbrush with a reciprocating motion or even a <cough> vibrator be more effective?

IC Diamond is VERY difficult to get to spread particularly on a cooler surface that has microgrooves from machining on it.

 

[Rudy Toody]

It needs to be an up-and-down motion on the top of the heat-sink. The motion of the beard trimmer (or any hair clipper) on its side with the cutter perpendicular to the plane of the chip will do it. I suppose a <cough> vibrator with the pointy end pointing down on the heatsink will do it.

The possible problem with using the buzz, is that if the vibrations are too strong, they could perhaps loosen some connections on the mobo. I have use is successfully on six different rigs and mobos without a problem.

 

[cboath]

Never would have thought of the beard trimmer technique

Thanks to all for the discussion, too, i find it very very interesting. I asked originally because after setting my machine up over the weekend and oc'ing to just 3.2ghz - i ran prime and after 10ish minutes the temps were 74-77 across the cores which seemed pretty high from what i had been reading. I'm not overly knowledgeable on oc'ing, esp on the i7. I redid a few settings (hardcoded some voltages mainly) and reran the prime test. After 2-3 minutes, the temps stablized and held to about 64 max from there until i stopped the test at about 15 minutes. Big difference. I upped it to 3.4 and got 71 68 68 68 after 15 minutes. Still nearly 8 degrees cooler than before. So it doesn't really seem to be a paste issue after all. It's been commented that maybe the paste needed time to settle or something like that.

I still find this super informative and if I ever need to reseat the HSF, i think i'll be going like method. Guess i'll check the AS site for instructions on that. I have a noctua 1366 and it just said use the pea sized method.