Does Lapping a CPU / Heatsink actually drop the temp?

Question

i've been watching some YouTube vids about Lapping a CPU. I've never heard of this modding technique before and, though extreame, I was wondering if it acutally works?

Assuming you lap your cpu and/or heatsink correctly, will the temps drop? When I say drop, at least a 1 degree drop is success (for the debate of this topic).

To keep this topic clean, please refrain from anyone commenting on the overkill of labour, just for a 1 degree (worst case) drop, etc. This is a discussion about the theory and concept, not personal opionion of wether to lap or not.

Answer 1

As far as I can tell, there is only anecdotal evidence for this, something which is very dangerouse since those who broke their processors or actually got a temperature increase are likely to assume they did it wrong and not write about it. I have yet to see any clear evidence that this works but I also have not seen any clear evidence that it does not except ramblings about how "it could not possibly work, just because".

As for anecdotal evidence, on the local swedish overclocking forums the reported results where a decrease was reported seem to vary around 2-5 degrees.

Also, there is a very good chance that the processor/heatsink is very smooth and flat to begin with and that this will actually result in a higher temperature.

If you're going to try it I suggest starting with the heatsink since there is practically no chance of breaking it, while there is a very real chance of making your processor into a very expensive keyring.

UPDATE: As hyperslug says, there is some compelling evidence in these two posts that supports the tweak as a significant improvement. (~10C difference)

Answer 2

I would say: Yes, it does work. I can't fathom a reason of why heat transfer wouldn't improve if one/both surfaces have been polished up. (Done properly)

But before you start lapping you should review what you are trying to gain from it. I have read of experiences where (done properly) has ranged from 2-8 degrees of improvement. Is that amount of temp drop worth the effort? Don't lap your CPU if you are doubting yourself. The heat sink should be enough for a result.

If you are still interested have a look here. It is full of instructions and nice pictures of lapping.

For everything else there is Google.

Answer 3

Years ago when the quality of the CPU caps were definitely in question (surface variances) this did matter. Reduction in temps of 8-10C though is not realistic for most users today (2012).

Also, the reasons for doing this were born from the fact we had to use stock heatsinks which were woefully inadequate of handling any extreme changes in processor temperatures due to overclocking. 80-92mm heatsink fans were enough to handle stock settings but not much more than that. Since then, the quality of the caps are much higher than they ever were AND we have much better choices now in cooling the CPU that were unheard of just a few years ago.

Watercooling, sealed watercooling units, extreme pipelines, push/pull stand up heatsinks, 120mm fans (single and or dual config) these have all made it practically useless to bother with lapping anymore. If at best you're going to get a "probable" 1-3 degree C temp change, is it worth 2-3 hours or more of work? And almost all users who lap don't even start with a known setpoint so they can't even tell what, if any gains, they've actually made. These are not assumptions they're known facts.