Does CPU cooler mounting pressure really matter?

[orangecat]

So I have a 3570K and as most people know Ivy Bridge chips had a huge issue with heat because of the crappy TIM used below the IHS. My chip used to run well above the 90°C mark when overclocked and stability testing so I delidded my CPU a while back and when I did I used that TIM Noctua includes with all their CPU coolers and my temps were still high. I eventually got fed up with it and decided it was worth trying to reapply it and when I did I saw nearly a 30°C drop in temps. Now I'm not sure if I poorly applied the TIM the first time or poorly mounted the cooler but it got me thinking... how important is CPU mounting pressure anyways? I've heard that you need to wrench it on there and I've also heard as long as its making contact it's fine. I think I put the cooler on tighter this time around and I noticed all of my CPU cores had an improvement in temps not just 2 like it was before.

 

I don't remember if LMG made a video about this but if not they should.

[zMeul]

over-tightening the cooler, if the mounting system doesn't have stops, can crush the LGA  socket pins

 

 

it was in issue with 1151 (Skylake) for users who used non spec coolers - it crushed the pins

[Canada EH]

Yes do not over tighten

[RadiatingLight]

I'd say go with medium pressure. go until you encounter resistance with all 4 screws, and then go a tiny bit farther than that (like 1 turn)

[WoodenMarker]

You're encountering pump out issues caused by heat cycling. The TIM Intel uses under the IHS is excellent and is designed to not encounter this problem. The reason why temps are lower after delidding is because of the gap being lessened that was caused by the adhesive under the ihs and around the die.

NT-H1 will usually work for ~2 weeks depending on your usage until it starts to not work as well or even worse than pre-delid. The solution is to use something that doesn't pump out like liquid metal. I personally use CLU. 

[ThomasD]

9 hours ago, WoodenMarker said:

You're encountering pump out issues caused by heat cycling. The TIM Intel uses under the IHS is excellent and is designed to not encounter this problem. The reason why temps are lower after delidding is because of the gap being lessened that was caused by the adhesive under the ihs and around the die.

NT-H1 will usually work for ~2 weeks depending on your usage until it starts to not work as well or even worse than pre-delid. The solution is to use something that doesn't pump out like liquid metal. I personally use CLU. 

Don't really have enough experience to speak with authority but that sure does make sense.  The liquid metals truly do wet the surfaces, so are going to be held in place by surface tension - which is a surprisingly strong phenomenon.  Meaning that even if thermal expansion wants to force the material out of the gap it will tend to return upon cooling.

 

Thinking about it logically I doubt the amount of mounting pressure matters much - so long as it is sufficient to create and maintain an even 'sandwich' of TIM between the two solid surfaces performance should be no different than if the pressure was something greater.

 

Edit:  Although it should be noted we are talking about the amount of pressure - not the thickness of the TIM layer.  A thicker layer will tend to perform less effectively than a thinner layer simply because the heat has further distance to traverse.