Applying CPU Thermal paste: X-method??

Haha I actually got my CPU from Andrew “dRweEZ” Roberts. Sadly it didn't have what it takes and was binned, so got it for a good price.

He did a wonderful lap on it though. It's been running very happy under water for around 2 years ..
 
Last edited:
Which thermal paste is recommended?

I used to use Arctic Silver before, but that was years ago.. I checked with Frontosa yesterday and doesn't look like they stock it anymore, they now only keep Coolermaster.

They have a few different types in the CM range, but I'm note sure which one to use.. It's for a Dell XPS Notebook.
 
mh348 said:
Which thermal paste is recommended?

I used to use Arctic Silver before, but that was years ago.. I checked with Frontosa yesterday and doesn't look like they stock it anymore, they now only keep Coolermaster.

They have a few different types in the CM range, but I'm note sure which one to use.. It's for a Dell XPS Notebook.
Click to expand...

There was similar questions asked not too long ago .
SEE THIS THREAD :)
 
1441568916 said:
You are wrong. However let's just leave it at that, I don't have the time or cash on hand to prove you wrong, at this time of year.
Click to expand...

Actually Archer is correct.

Metal to metal gives the best thermal transfer. A block of copper will transfer heat from one side to the other quite effectively so it's not metal that's at fault.
The problem is that even a nicely machined metal surface has microscopic hollows/pits in it. These pits prevent the metal surfaces from contacting each other perfectly and air is trapped in these pits which happens to be a great thermal insulator.
What thermal paste does is it fills up the pits and thereby displaces the air which would otherwise be trapped and this is why thermal paste helps increase thermal conductivity.
 
Soul Assassin said:
I know this is the internet but I wouldn't lie to you, you can just Google CPU delid results, the numbers are there.
Click to expand...

While this thread is revived - the topic of delidding:

[video=youtube_share;LvS_qgn7OSs]https://youtu.be/LvS_qgn7OSs[/video]

I still believe a 20C drop in temps is absolute bs unless there was something wrong with that CPU from the factory, this video seems to support the idea that the gains are minimal vs the risk.
 
DrJohnZoidberg said:
I call bs on that number. Why would Intel, the largest desktop CPU manufacturer, not already just be doing this? 20C is a massive number! I'd maybe believe it if you said 2C maybe.
Click to expand...

Because they save money on the thermal paste, and their production process for putting on the heatspreader along with the glue is just good enough to ensure that most chips all have the same thermal characteristics out of the box. Removing and reseating the heatspreader takes away some of the air gap underneath, and facilitates better heat transfer. This has become more difficult to do with Broadwell and Skylake chips though, as Intel has transistors and other stuff up on the top of the packaging now, so there's a greater risk of damaging those while taking off the heatspreader.

DrJohnZoidberg said:
While this thread is revived - the topic of delidding:

[video=youtube_share;LvS_qgn7OSs]https://youtu.be/LvS_qgn7OSs[/video]

I still believe a 20C drop in temps is absolute bs unless there was something wrong with that CPU from the factory, this video seems to support the idea that the gains are minimal vs the risk.
Click to expand...

I expected Linus to be pretentious about his findings. He might be right about the risk, but he's wrong about the benefit.

If they had put Luke on this topic, he would have mentioned that Haswell owners saw big drops in temperature from doing this because Intel put the voltage regulator onto the CPU package, which resulted in much, much higher temperatures. Delidding is something that any self-respecting pro overclocker also has to do in order to obtain higher scores on air or water. Sometimes putting the heatsink into contact with the die itself also yields better temperatures.

Also, on Broadwell-C chips, you have to run your heatpaste on one side of the heatspreader because that's where the CPU die is. Delidding Broadwell also has temperature and slight performance benefits because of this.

Edit: Also, don't just take my word for it. There's a huge community of enthusiasts out there doing this, and it's not for no benefit either.
 
Last edited:
Linus is smart, but the way he treats hardware sometimes makes me cringe. He destroyed so many boards trying to make that PC router work. T_T
 
yes, i feel this whole topic as a whole in the PC industry is a waste of time. There is no difference between these methods, and if there is, it is very negligible...
 
ariz04 said:
yes, i feel this whole topic as a whole in the PC industry is a waste of time. There is no difference between these methods, and if there is, it is very negligible...
Click to expand...
No, is not. Removing radiator usually shows up a thick layer of dried-out compund and quite large completely empty space not covered by the paste at all. It means that some portions of CPU structure are overheated.

The most reliable method is to spread paste evenly across the entire surface (to both surfaces) and remove excess with a plactic card. Then attach CPU and rotate little-bit in both directions under pressure. No X-method or a drop is acceptable in my records, they create a thick layer and empty spaces.
 
sajunky said:
No, is not. Removing radiator usually shows up a thick layer of dried-out compund and quite large completely empty space not covered by the paste at all. It means that some portions of CPU structure are overheated.

The most reliable method is to spread paste evenly across the entire surface (to both surfaces) and remove excess with a plactic card. Then attach CPU and rotate little-bit in both directions under pressure. No X-method or a drop is acceptable in my records, they create a thick layer and empty spaces.
Click to expand...

The proof is in the pudding though. It shows there is no difference in temps what so ever. So do I use a couple of minutes trying to apply it equally with a card and wiping of the excess, or do I just put a pea size in the middle? The force exerted on the CPU when you put the Cooler is more than enough to spread it out equally. Anyway, whatever works for you guys. There is no evidence to prove otherwise.
 
Last edited:
sajunky said:
No, is not. Removing radiator usually shows up a thick layer of dried-out compund and quite large completely empty space not covered by the paste at all. It means that some portions of CPU structure are overheated.

The most reliable method is to spread paste evenly across the entire surface (to both surfaces) and remove excess with a plactic card. Then attach CPU and rotate little-bit in both directions under pressure. No X-method or a drop is acceptable in my records, they create a thick layer and empty spaces.
Click to expand...

Check the video I posted. No difference
 
I know you always disagree in this matter, but I removed many heatsinks and from my experience a pea of paste gives the worst results, both create a very thick layer comparing to the method I use.
 
sajunky said:
I know you always disagree in this matter, but I removed many heatsinks and from my experience a pea of paste gives the worst results, both create a very thick layer comparing to the method I use.
Click to expand...

I agree, find the best results are with a thin spread using credit card.
 
Archer said:
That is not quite correct
Metal on metal contact is far superior to any paste out there. The paste is there to fill the holes where metal on metal contact is not occurring, ie where there would have been metal, then air, then metal.
Click to expand...
You are absolutely correct. By far the best for heat transfer is metal-to-metal contact between the CPU die cover and the heatsink. But in mass manufacturing the surfaces have imperfections and so the two don't match up perfectly. Thermal Interface Material (TIM) is used to fill the microscopic air gaps between the two surfaces. A widespread misconception is that the TIM is better at heat thermal transfer. This leads to the even more widespread mistake of using too much TIM. The optimal way is to have the TIM fill only the microscopic air gaps where the metal-metal surfaces do not make perfect contact.

Edit: Oh, I see Paul answered this already.
 
Last edited:
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X