Question Thermal paste gone wrong

[hardware_bxl]

The cooler is Noctua NH-D15 Chromax, the CPU is AMD 7900x : after I removed the cooler and checked the thermal paste (see the link for photos), I noticed the thermal paste was all to one side of the cooler's contact plate, I guess this is not good. Should I change the offset for this cooler and how would I do this? or is this normal?
It's my first build, so not sure.
Also, I know it's not critically important, but the thermal paste does not look like it covers the cpu properly, now that I will re-do the thermal paste, how to do this better?
The thermal paste I used is Thermal Grizzly Kryonaut as I was told this was a good choice, but I also have thermal paste that came with the Noctua cooler.
Which thermal paste to use and how to apply it best?

Thank you all so much for your help! First time builder's appreciation is huge ;-)

Here are the images

 

[hardware_bxl]

I mean, 70C vs 98C is quite the difference, I could easily accept 80, maybe even 85C if I hear you have max 70C spikes. Get me? I can remove the cooler again and check the thermal paste, replace it with mx-4, apply a thin layer and all that, but all that is never going to get the temps down to anywhere below 90C i am sure.
So annoying... Idle temps are fine, it's not about that, if you idle at high 30's, then my 44C is quite acceptable.. and I am using the 7mm offset option with the noctua bracket.. 38-39 vs 44 in idle, ok.. but 70 vs 98 i don't understand what i am doing wrong, or if it's just my chip that is worse quality.

 

[bill001g]

Take the side panel off the computer and see if it makes any difference. Make sure the fans are running at 100% so you know the cooler is doing all it possibly can.

You also have to be careful about comparing numbers. From time to time we see people posting from india on this forum and running computers in houses with little air conditioning. There actually is a huge difference in results if you compare running the same computer in a house at 25c to someone at 30 or even 35c.

 

[JayGau]

I can't believe that in 2025, with the dozens of videos available online, there are still people who think that "too much paste is bad" and "you should not put more than a small pea". This is a myth. All the tests show that you need to put enough so the whole IHS will be covered once the pressure gets applied. Too little is bad, not the opposite. There is no such a thing as too much paste, in a cooling point of view. The only downside of way too much is the mess the overflow can do on the mobo.

Several tests showed that the best CPU temperature is achieved with the "cross" application method, but it's only because this method requires more paste than the others (you can't draw the cross with an amount the size of a small pea). And in every single test, the worst result is always with "not enough paste", i. e. the small pea that so many people still believe is the right amount.

 

[hardware_bxl]

@JayGau : I kind of knew this too, but just preferred to try and do it the 'optimal' way.
I am not even worried about the thermal paste that much right now, but I am kind of annoyed by the high temp differences between same cpu's doing an out of the box standard Cinebench 2024 multi-core test.
@bill001g : I know there are people posting different numbers from other locations, but for this post I am only referring to a guy from Ireland with a similar situation and his 9800x3d is maxing out on 70C even when Cinebench runs for an hour+, compared to mine who reaches 95C+ after 5 minutes of this same test.
I know he uses AIO cooling with 360 rad, but even then, the NH-D15 cannot be the reason imo for this difference.
For a same cpu, you would expect 5-10 max degree difference no? At least, in normal conditions.
So that part I am still just worried about since I have no prior experiences building a system, that I have done something wrong, but I just cannot see what part.

 

[hardware_bxl]

I will test later today with side panels off and 100% fan speeds for all fans, but I cannot imagine this bringing the peaks for the 9800x3d down dramatically, but i will give it a try.

 

[MEMOFLEX]

That makes me just sceptic, not because of what you are saying, but that would basically imply that either I did a very bad job with my installation, or the temperatures differ so much from chip to chip, that would almost make me want to return my 9800x3d, but I am sure they will deny the RMA...
I know you have an AIO installed, but the Noctua cooler is claiming to be one of the best air coolers, so that would still not make me happy seeing these 98C temps at full multi-core load... :-(

Prior to applying the -15 all core negative offset my temps were around the 85 mark for a sustained load in cinebench. Have you tried applying a negative offset to see if it helps with the temps? Sometimes the motherboard can apply overly aggressive settings so this might be worth looking at too

I do have a very well ventilated case (positive pressure) and my corsair h150i titan cooler is also in a push pull configuration bringing in fresh air from outside the case. I moved my AIO from exhaust to intake which helped the cpu temps.

I feel it is maybe a combination of all these things plus a little bit of silicone lottery coming into play.

 

[geofelt]

Noctua will have instructions on how to apply paste.
They know what they are doing, I would follow that method.
The pea method used to be a good one, but I think the directions for AMD have changed.

Back to basics:
The purpose of paste is to fill in microscopic pits in the mating surfaces.
Air in the pits is a poor conductor of heat.
Metal to metal contact is best, but paste conducts heat better than air.
Hence the admonition to not use too much paste.
Only enough to fill in the pits, but leaving as much metal to metal contact as possible.
It is possible that a manufacturing defect has left one of the mating surfaces uneven, but, I think this is unlikely.

On the first couple of images, it looked to me like the cooler was not tightened down evenly.
That seems hard to do with the noctua mount.
Perhaps one side was not screwed all the way down to the stop.