Core 0 and 2 hotter than 1 and 3 - location on CPU?

[Tan3l6]

Hi!

 

I just replaced my CPU cooler. 

The problem is - 2 cores are hotter than other 2

 

Is there some indiocation where the 0 and 2 are located physically on CPU? So I'd know which side has bad contact.

 

The core temps are basically the same on the two pairs.

[WillLTT]

1 minute ago, Tan3l6 said:

Hi!

 

I just replaced my CPU cooler. 

The problem is - 2 cores are hotter than other 2

 

Is there some indiocation where the 0 and 2 are located physically on CPU? So I'd know which side has bad contact.

 

The core temps are basically the same on the two pairs.

thermal paste not covering everything?

[Skiiwee29]

There will usually be some disparity between the core temps, usually in the 5c range on average.. if its greater than 10c, thats usually an indication of other issues with either the TIM between the die and IHS, the TIM between the IHS and cooler, IHS flatness etc... 

[BigRom]

1 minute ago, Tan3l6 said:

Hi!

 

I just replaced my CPU cooler. 

The problem is - 2 cores are hotter than other 2

 

Is there some indiocation where the 0 and 2 are located physically on CPU? So I'd know which side has bad contact.

 

The core temps are basically the same on the two pairs.

How much hotter, if the temperature difference isn't too big (less than 5C) I wouldn't really concern myself with it as not every core is equal, like some cores are able to boost 1-2% higher than others

[Tan3l6]

Yeah, the difference is big ... on some Prime95 load 70C versus 57C

It's a real hassle to take it apart and put it together just to do it again. So maybe there could be some prediction - what side has cores 0 and 2?

[Tan3l6]

The CPU is delidded and has liquid metal between IHS, not fixed though. But my previous cooler was the same - not glued back.

Thermalö paste is covering all, pretty sure.

 

I already put the cooling on and off, succeeded to make computer boot in second try. (Had 5 long beeps and nothing else)

[Skiiwee29]

1 minute ago, Tan3l6 said:

The CPU is delidded and has liquid metal between IHS, not fixed though. But my previous cooler was the same - not glued back.

Thermalö paste is covering all, pretty sure.

 

I already put the cooling on and off, succeeded to make computer boot in second try. (Had 5 long beeps and nothing else)

Then sounds like the issue is your IHS isnt flat causing issues. Try lapping your IHS to flatten it out and this will likely resolve it. 

[Tan3l6]

Damn. You might be right. The previous cooler had the same IHS and liquid metal and temp difference between cores was about 2-3 degrees max though.

Might be CPU core and IHS bad contact.

 

Too bad I haven't got a workbench to test the temps before putting it all together.

[Skiiwee29]

Here is an example of the difference a lapped IHS can make.

 

 

[Tan3l6]

Well I just reapplied liquid metal and thermal paste and the difference is lower, though one core is still noticably hotter than others. I still wonder where is core 0 ... but IHS should be lapped, agree.

Well, got the cores under control:

 

[Vivid Dream]

What did you do to get your temps under control?

[Falkentyne]

On 10/6/2019 at 4:32 AM, Tan3l6 said:

Hi!

 

I just replaced my CPU cooler. 

The problem is - 2 cores are hotter than other 2

 

Is there some indiocation where the 0 and 2 are located physically on CPU? So I'd know which side has bad contact.

 

The core temps are basically the same on the two pairs.

 

The cores are labeled like this.

This is for 9900k, but 4 and 6 cores follow the exact same method.  
(HWinfo based numbering):

 

 

 

The cpu is convex, and this usually causes the top cores to be hotter than the bottom cores.

Note: i haven't figured out why its always the top cores.  Why not the top cores cooler and the bottoms hotter?  Guess that's under Intel's secrets.

Sanding the IHS can help, but sanding the IHS doesn't really fix the problem with the die to IHS contact balance at all; but it does improve what's there since the IHS is also going to probably be slightly convex (hoping the heatsink then is slightly concave, in which case both must be lapped).

 

The obvious fix is to sand all three slightly..bare die, IHS and heatsink.  But well you know where that leads...

 

People who delid these processors and go direct die cooling still have the same core temp deltas.  Seen it on bare die 9900K's.  Just the deltas wind up being much closer than they were before.

 

BTW: a HUGE IMPORTANT protip to anyone delidding any processor and using liquid metal:

 

DO NOT polish any surface to a mirror finish that will accept liquid metal !  I repeat--do NOT.  That includes the CPU die also.  

Liquid metal needs a slghtly rough, but flat surface to maintain surface tension.  If you have some sandpaper and a spare flat piece of copper, nickel or some other non aluminum heatsink metal, you can see this in real time for yourself.  Try sanding something to mirror with up to 7000 grit sandpaper then watch what happens when you try to apply a thin coat of LM and spread it out.  You will see what happens to the areas where the LM is thin at.

 

What you need to do is take some 1500 to 2000 grit dry sandpaper and buff the surface of the IHS and CPU die to roughen it up (you don't need glass for this, just your fingers will do).  Then wipe and clean it with a lint free cloth and apply LM on the roughened surface.  You will see a huge difference in how the LM spreads.  (the same thing should be done to heatsink and top IHS for those who apply LM on top of the IHS, instead of normal paste  RIP your warranty though :):)  Instead of pooling up, it will spread slowly and even have a different look to it.  And if it's in a very thin layer after pressing (softly) with the spreader, it will no longer leave exposed spots on the surface from clumping together when thinned.  Also make sure you don't press too hard and there are no resistances from 'micro specs'.  Micro specs can happen if you apply too much pressure when applying LM (coming from the surface itself) and any tiny little particle will have a terrible impact on temps.

 

Mirror finishes work great for normal thermal paste.  Not for Galinstan !

[Tan3l6]

3 hours ago, Vivid Dream said:

What did you do to get your temps under control?

I removed liquid metal under ihs and carefully applied it again, also reseated cooler few times.

Every time I had to remove the motherboard because of CPU cooler attaching design.