Their conclusion was that nearly any substance can be used that properly fills the space under the heat sink. It is better to use a substance that is not perishable and has a long working life. The expensive thermal grease performed no better than the inexpensive common variety with equally probable lifetimes. The rest is all marketing and old-wives-tales.

I've seen something similar but can't remember where. Though it may be true there's little difference in results for short term use, there's likely a difference in long-term performance. Then there's the matter of changing a processor when a compound has hardened. These factors should be considered when selecting a thermal compound.

---

Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

An interesting test was done by a prominent hardware review organization of the actual performance of a variety of heat transfer compounds. I don't remember who this organization was but I found their report on the internet recently and I believe the weren't lying. They tested a variety of possible heat transfer compounds including these:

artic silver
ordinary thermal grease
peanut butter
vegamite
mayonaise
a heat transfer 'pad'

Yes, they did! With the exception of the heat pad which had poorer performance, no difference in heat transfer efficiency was observed!

Their conclusion was that nearly any substance can be used that properly fills the space under the heat sink. It is better to use a substance that is not perishable and has a long working life. The expensive thermal grease performed no better than the inexpensive common variety with equally probable lifetimes. The rest is all marketing and old-wives-tales.

I'm thinking of trying peanut butter and jelly next time I take my machine apart. Or maybe ham and cheese.

Hi,

I have tried lapping on a heatsink for an Athlon 2200+ and it worked really well. In my case temps were around 3 degrees lower after lapping. I used 400, 600 and 1000 grit paper and then finished with steel polish until I could see my reflection quite clearly. It took me an hour or so.

How much lapping brings will in the first place depend on the quality of your heatsink's contact surface: the smoother it is the less will additional lapping bring. [ "Better" is not to be mixed up with "more expensive", however. I have seen poorly lapped heatsinks in the highest price categories and good ones in low-cost. ]

And finally: cotton with thermal paste? Yuck... you might as well try to spread honey on your bread evenly with a brush with loose hairs. Anything that lies between your heatsink and CPU is a killer (even too thick thermal paste...).

Regards,

Christoph

---

"I know that you believe you understand what you think I said, but I'm not sure you realize that what you heard is not what I meant." R.M. Nixon

... the suggestion to use a cotton swab to apply the thermal grease doesn't seem too smart.

Putting it mildly.

Trapped cotton fibers could interfere with heat transfer.

Definately would.
A cotton swap & some acetone would be good for removing old grease, but i wouldn't use a swap to apply any.

---

Grant
Darwin NT

I've been doing some searching on the net trying to get as much information as I can about how to properly apply thermal compounds onto CPUs, and the information varies from site to site. But I did find a fairly good article at Hardware Secrets about the subject: How To Correctly Apply Thermal Grease

I'm going through this out of curiosity but the suggestion to use a cotton swab to apply the thermal grease doesn't seem too smart. Trapped cotton fibers could interfere with heat transfer.

---

Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

I've been doing some searching on the net trying to get as much information as I can about how to properly apply thermal compounds onto CPUs, and the information varies from site to site. But I did find a fairly good article at Hardware Secrets about the subject: How To Correctly Apply Thermal Grease

@ tekwyzrd. Your mention of lapping the heatsink made me search for some info on that cooling method as well. I found some lapping before and after pictures at Overclockers Club. I can see what difference that might make for heat transfer but I don't have any experience actually doing it though, so I'm afraid I'll probly screw it up. But thanks for the tip.

---

I believe that AS5 is supposed reach peak efficiency after several cycles of heating and complete cooling, rather than a single extended period of medium or high heating.
As Grant says there may not be much change after break-in, but every bit can help.

I see. So your saying to run at load for a little, then turn off the computer to cool, and just keep repeating that? I can try that. It's what I did last night, I'll keep that up.

Lapping can make a difference too. The smoother the surfaces the better.

---

Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

I believe that AS5 is supposed reach peak efficiency after several cycles of heating and complete cooling, rather than a single extended period of medium or high heating.
As Grant says there may not be much change after break-in, but every bit can help.

I see. So your saying to run at load for a little, then turn off the computer to cool, and just keep repeating that? I can try that. It's what I did last night, I'll keep that up.

---

Only thing I'm wondering now is if I should run the processor at maximum or wait 200 hours with the processor running at normal levels for the compound to break-in.

Go for it.
The "break-in" period may result in ½° reduction in temperature. Maybe.

I ran the processor at max for about an hour last night, then shut it down to cool a little. Well I guess your right, the load temp is stable now at 69°, where before it was at 70° with occasional reduction to 69°. So it's about a 1/2° reduction so far.

---

I believe that AS5 is supposed reach peak efficiency after several cycles of heating and complete cooling, rather than a single extended period of medium or high heating.
As Grant says there may not be much change after break-in, but every bit can help.

---

My Dell Latitude D600 ( Pentium-M 1.6GHz ) with Artic Silver 5 is running idle @ 43o. and FULL Load with Seti @ :

58o. - Slow Fan Speed
53o. - Highest Fan Speed

Best Regards,

Rafael Valente

---

Only thing I'm wondering now is if I should run the processor at maximum or wait 200 hours with the processor running at normal levels for the compound to break-in.

Go for it.
The "break-in" period may result in ½° reduction in temperature. Maybe.

---

Grant
Darwin NT

Okay, after a little surgery, I went ahead and removed the previous AS-5 compound, cleansed and purified the contact surfaces with ArctiClean 1 and 2, and re-applied a new layer of AS-5 that's about the size of a BB. Only thing I'm wondering now is if I should run the processor at maximum or wait 200 hours with the processor running at normal levels for the compound to break-in.

Thanks Grant (SSSF) and hazmatt87 for the input. :)

---

The card spreading method works best on exposed cores like on video cards. I prefer to use a razor blade when doing it that way.

Exposed core or heatspreader- the theory is still the same.
The heatsink should be in contact with the core/heatspreader. The heatsink compound is there to fill in the microscopic air gaps between the 2 surfaces. Having too much results in poor thermal conduction, even more so in the case of a large surface area ie heatspreader.

---

Grant
Darwin NT

I wouldnt recommend spreading the AS-5 over the heatspreader. You always wind up with too much in between the surfaces and theres also a greater chance of bubbles doing it that way too. All my best results are using just a little dot of AS-5 in the middle and using my waterblock (or HSF in your case) to smoosh it down and thin it out. Keep in mind that only a very thin layer of TIM is needed to effectively transfer the heat, too much and it can deter the conduction. Only use more if your surfaces are in bad shape (scratched up). The right amount you need is about the size of a BB.

Hmmm, that's what I'm thinking, using the size of a BB instead of 1/2 a BB like Arctic Silver says. I guess the heatsink pressure will still thin it out, only it'll cover more of the IHS/heatsink contact area. I should mention that the previous less quality compound that was on was spread out throughout the IHS and you could see it comming down the sides a little. The maximum temperature when running BOINC then was 68°.

You may want to consider a cooling upgrade, there are a lot better aftermarket heatsinks around than the stock prescott cooler. You on socket 775 or 478?

The cooling is fairly good for a notebook. It has 2 50mm fans and 3 copper heat pipes, non-Intel.

It's a 532 mobile Prescott socket 478.

---

I wouldnt recommend spreading the AS-5 over the heatspreader. You always wind up with too much in between the surfaces and theres also a greater chance of bubbles doing it that way too. All my best results are using just a little dot of AS-5 in the middle and using my waterblock (or HSF in your case) to smoosh it down and thin it out. Keep in mind that only a very thin layer of TIM is needed to effectively transfer the heat, too much and it can deter the conduction. Only use more if your surfaces are in bad shape (scratched up). The right amount you need is about the size of a BB.

The card spreading method works best on exposed cores like on video cards. I prefer to use a razor blade when doing it that way.

You may want to consider a cooling upgrade, there are a lot better aftermarket heatsinks around than the stock prescott cooler. You on socket 775 or 478?

---

To me it makes sense that the thermal compound should be between the entire CPU/heatsink contact area and not just at the center.

Yep.
But covering the whole contact area should be done by smearing the heasink compound all over the area in question & then scraping it off using a credit card or busniness card.
Having a smear over the whole thing results in poorer heat conduction. The compound is meant to fill in the small gaps between the heatsink & the CPU surface- it is not meant to fill in the entire space.

Thanks for your response.

I understand that the heatsink should do the smearing and it should be directly over the compound during placement. When I applied the new compound, I only added a very small amount at the center and put the heatsink right over it. It's supposed to spread out over the next 200 hours, but I don't think the small amount is doing the job. Are you saying that I should just add slightly more at the center of the IHS, lower the heatsink, and maybe it'll cover more of the IHS, or spread a small amount over the entire IHS? I thought that the card-spreading method worked better for open cores instead of CPUs with integrated heatsinks.

---

To me it makes sense that the thermal compound should be between the entire CPU/heatsink contact area and not just at the center.

Yep.
But covering the whole contact area should be done by smearing the heasink compound all over the area in question & then scraping it off using a credit card or busniness card.
Having a smear over the whole thing results in poorer heat conduction. The compound is meant to fill in the small gaps between the heatsink & the CPU surface- it is not meant to fill in the entire space.

---

Grant
Darwin NT

When running BOINC my processor is getting a little too close to its maximum thermal operating zone. I'm using a new notebook computer with a Mobile Pentium 4 532 3.06GHz Prescott and it sustains a temp of 70° celsius when running BOINC. The maximum operating temp is 75°, so you can see what my problem is.

A couple days ago I replaced the standard OEM's thermal compound with Arctic Silver 5. The P4 532 has an Integrated Heat Spreader (IHS), and Arctic Silver says to place only a tiny amount at the center of the IHS, so that's what I did. That may be my problem. I know how Arctic Siver works with exposed cores, but does anyone have an opinion about wether or not it should actually be spread over the entire heat spreader? To me it makes sense that the thermal compound should be between the entire CPU/heatsink contact area and not just at the center.

---