I have a Lenovo ThinkPad P1 20MD with a i5-8400H quadcore CPU. Base clock is 2.5 GHz, turbo clock is 4.2 GHz for 1-3c, and 4.1 GHz for 4c.
As of lately (maybe 2 weeks or so?) I've been noticing that whenever I run CPU-intensive apps or benchmarks, the CPU temperatures do not get distributed evenly across cores. In fact, they seem to always follow a particular pattern: 1st core is always the hottest, followed by 2nd, then 3rd, then 4th core. The temperature difference between the 1st and 4th core is sometimes as low as 2C, and sometimes as high as 10C. My PROCHOT is set to 98C, so when the 1st core hits 98C, the 4th core is typically around 92C.
I understand that this may not be an atypical core temperature delta for Intel CPUs, but I'm concerned that it's following a distinct pattern.
Am I correct in thinking all signs point to a poorly seated heatsink that needs to be repasted and reseated? Or could something else be going on?
Screenshots:
HwInfo
ThrottleStop
![]()
![]()
-
CitizenInsomniac Notebook Enthusiast
-
Could be the heatsink, but I doubt it. I've noticed this for over a decade now. I don't even remember Core 2 Duo now (I think it also had the same , but Core 2 Quad could be explained away by it having 2 separate dies. However all of my i7 have had the same exact behaviour.
And of course I tried reseating and lapping the heatsinks - never made a difference. Always a difference in temperature, 2-10 degrees as you have noticed.
Since I've not had any success with adjusting the heatsink and no errors or failures, I've come to accept that's just how they behave. It's probably just the internal sensor being closer to whatever core is hottest. Or a combination of that and a very slightly uneven heat dissipation (as heatsink-die aligment can never be 100% perfect).
Desktop Intel processors with integrated heatspreaders have the exact same behaviour, you can find countless threads online with people worrying about the same thing since multicore CPUs came to marketCitizenInsomniac likes this. -
CitizenInsomniac likes this.
-
CitizenInsomniac Notebook Enthusiast
I guess repasting the CPU couldn't hurt, right? Worst case scenario, it stays the same, but best case scenario - it improves both the heat dissipation and distribution. Or are there any risks I am not considering? -
If you've never done that, it will very likely improve your temperatures. Just don't damage anything, y'know
-
-
CitizenInsomniac Notebook Enthusiast
Quick update:
I repasted my CPU & heatsink with Phobya NanoGrease Extreme, and the initial results (at least from a thermal dissipation perspective) are good. CPU core temps idle around 42C on AC power, and around 35C on battery power (I've also seen them go as low as 30C). When running CPU stress tests the temps seem to take a little longer to get to 98C (my PROCHOT), but ultimately they still reach 98C and then thermal throttling kicks in. All in all, it seems like an improvement, and I've gained about 100 points in Cinebench R20 (new high score: 2138 multi; 434 single).
The only trouble is... The temps are still distributed unevenly across cores, maybe even more so than before. When the temps are high, the delta between the hottest and coolest core is now around 10C (e.g. 95C vs 85C).
So unless I've had the bad luck of misapplying thermal paste in exactly the same uneven way as they did at the factory, I'm going to go ahead and guess that it's not the paste that's causing this.
If that's just how it is with Intel CPUs, OK, I'm fine with that. But is there any way to somehow redistribute the workload across CPU cores to compensate for it? For example, if my 1st core is at 88C when the 4th core is at 98C (PROCHOT), could the OS or CPU somehow prioritize the 4th core higher? It somehow feels like a waste that the entire CPU package gets throttled just because one core hit 98C. -
Papusan likes this.
CPU temperatures distributed unevenly across cores
Discussion in 'Hardware Components and Aftermarket Upgrades' started by CitizenInsomniac, Apr 24, 2020.