Question CPU temperature increasing over time

[Dave3000]

I have a Ryzen 5800x that I replaced the thermal paste with Arctic MX4 about 5 months ago. Back then it took about 25 seconds for the CPU package temperature to reach 90C and my Cinebench scores were higher. Now the CPU package temperature hit's 90C instantly and I'm experiencing more thermal throttling that before. 5 months ago when I replaced the thermal paste (Corsair XTM50) with new thermal paste (Arctic MX4) an it would score in upper 5900's to low 6000's in CB20. Then months passed and I was experiencing more throttling and CB20 scores now reach low 5800's to mid 5900's. Could it be the weight of the heatsink/fan assembly of my Noctua U12A overtime loosening up the contact with the CPU. I'm using one of the best thermal pastes around, so it can't be the thermal paste quality. I used the X-method, but I didn't find any difference in thermal paste application method for the nearly 2 years that I owned this CPU. If it's a weight issue (gravity pulling the heatsink downward), would an AIO be the solution to this issue? My case cooling is 2x120mm intakes and 1x120mm outake, but I also tried 2x140mm intakes and 1x120mm outake and my CPU actually ran hotter with the 2x140mm intakes than the 2x120mm intakes. I'm using a Fractal Meshify C case.

 

[DrMrLordX]

The mount pressure should be far greater in force than whatever is being caused by weight. It would help if you had voltage and power consumption numbers from five months ago for comparison.

 

[Kocicak]

Have you updated the bios in the meantime?

 

[ZGR]

Use HWiNFO to monitor your peak voltages. I will reach those temps past 1.3v, but we have different CPU's. My friend's 5800X reaches those temps reusing the air cooler his 3700X came with.

Also check inside BIOS, Ryzen Master, or PBO2Tuner for your PPT, TDC, and EDC. I highly suggest limiting it to 110/80/110 or somewhere around there.

I also recommend doing a negative voltage offset. I use PBO2Tuner to apply a -30mV per core undervolt and power limits as soon as I log into windows.

 

[Tech Junky]

Sounds like the paste is going bad.

I used a no name thermal grease on a laptop once and it worked great for about 4 days and then completely failed and temps shot to 100C. I went back cleaned it all off and put in some Arctic paste and it was good for a few months and I got sick of chasing temps and switched to a graphite pad on the CPU/GPU and it held steady after that. While the pad isn't the coolest it's the most consistent for me. I've been using the pad method on my server setup for a couple of years now though and prior to the laptop application and it's a no fuss / mess setup.

I picked up an ADL / 3060 laptop back in March and considering doing the swap on it as well as this thing puts out some serious heat under load. The temps don't quite hit the max but, I feel it could use some help and peace of mind.

 

[kognak]

Thermal cycles can squeeze paste out in certain conditions. Doesn't mean the paste is bad, just not good fit. I'd try something with higher viscosity, they tend to "stick" better. Though cooler removal after longer time needs to be done with care(warmed and with twists).

 

[Tech Junky]

considering doing the swap on it as well

Well I did it because I kept talking about it. The OEM paste was all dried up and flakey. one less thing to think about with it now.

 

[DrMrLordX]

MX-4 is kinda runny. There are better pastes out there, too. Personally I recommend Kryonaut for most HSF installations. It's good stuff. And it's thicc thick. It won't suffer from thermal pump-out.

 

[bononos]

I have a Ryzen 5800x that I replaced the thermal paste with Arctic MX4 about 5 months ago. Back then it took about 25 seconds for the CPU package temperature to reach 90C and my Cinebench scores were higher. Now the CPU package temperature hit's 90C instantly and I'm experiencing more thermal throttling that before. 5 months ago when I replaced the thermal paste (Corsair XTM50) with new thermal paste (Arctic MX4) an it would score in upper 5900's to low 6000's in CB20. Then months passed and I was experiencing more throttling and CB20 scores now reach low 5800's to mid 5900's. Could it be the weight of the heatsink/fan assembly of my Noctua U12A overtime loosening up the contact with the CPU. I'm using one of the best thermal pastes around, so it can't be the thermal paste quality......

Its not the weight of the heatsink, its probably a pump out problem. The popular thermal pastes like MX4 and HT1 have this issue especially with high heat density cpus which is probably all the newer cpus for many years now.
I'd advise using newer thermal pastes which are supposed to deal with this issue, they feel more viscous, look at the Mastergel Maker Nano, NT-H2, MX5, TF-X. The old MX2 I think is more viscous and some people say it has better long term performance.

Edit- I may be wrong about the pastes I mentioned. The NT-H2 is less viscous, none of the marketing materials for the newer pastes I mentioned said that they were specifically designed to combat pump out but the max operating temps for NT-H2 and MX-5 are higher than their predecessors. The MX-5 had a recall for an oil separation problem for a batch and the product is not listed on the Artic website which is a concern.

 

[bononos]

MX-4 is kinda runny. There are better pastes out there, too. Personally I recommend Kryonaut for most HSF installations. It's good stuff. And it's thicc thick. It won't suffer from thermal pump-out.

Kryonaut also suffers from pumpout, the newer Hydronaut product is supposed to to be better in dealing with this issue because it doesn't have silicone. TG's CEO DerBauer talked about pump out and Kryonaut/Hydronaut.
Generally all pastes except for some special cases like IC diamond will be more susceptible to pump out with higher temps and smaller dies (higher heat density).

 

[DrMrLordX]

Interesting. I used Kryonaut on my Radeon VII waterblock. If I ever remove the block, I'll have to check.

 

[bononos]

Interesting. I used Kryonaut on my Radeon VII waterblock. If I ever remove the block, I'll have to check.

If the temps are low, say <70C which is more likely on good water cooling, the pump out might still be manageable and repasting wouldn't have to be done that often. From what I've read, kryonaut seems to be susceptible to pump out like MX4/NT-HT1.

 

[DrMrLordX]

If the temps are low, say <70C which is more likely on good water cooling, the pump out might still be manageable and repasting wouldn't have to be done that often. From what I've read, kryonaut seems to be susceptible to pump out like MX4/NT-HT1.

I mostly run it undervolted so it doesn't get very hot. But! Radeon VII is infamous for hotspotting.