[SOLVED] 1 core temp is lower, 1 core temp is higher

[micbp]

hello everyone
just recently got my thermal paste replaced in the shop and noticed something about temperatures

on the idle the temperatures on all cores are pretty much the same (+/-1 degree)
but under stress, 1 core in particular seem to run hotter than the others (by around 5 degrees)
and other one, right below it, runs cooler than the others (by around 10 degrees)

so in between them there is a difference around 10-15 degrees under stress, which is bothering me
under whatever work cpu is doing, stress tests, gaming, the results are the same

1 core that is hotter, i know 5 degrees it is not that much, to be bothered in general
but that one that is cooler, makes me think that maybe thermal paste was applied pretty scarcely on all the others, except for that one spot?

i definitely dont remember such diffrenences before
using Throttlestop to check temperatures
i have i7-9750h cpu

please, correct me if im wrong in my suspicions

 

[jay32267]

I don't find it all that abnormal.
The sensors are sometimes not very precise.
If the temps are good I'd leave it be.
What are the max temps you see?

 

[logainofhades]

How flat the IHS is can affect core temps too. Hence why some pro overclockers lap theirs, to get a more even contact path, to the cooler.

 

[jay32267]

You could check the cooler mounting fasteners and make sure they are torqued evenly....maybe something came loose.

 

[Sohom]

actually u shud check the cooler placement...

 

[micbp]

its laptop guys, sorry, for not clarifying
everything should be fitted pretty good, otherwise it wouldnt close i imagine

so it is weird? am i correct?

@jay32267

Spoiler

max temps are at 100 C, but theyve been this way since purchase, after paste replacements, its just how this laptop colling is, whatever the cooling, it will just put on more power until it goes up to 100 and throttles. in any case im used to it, and its hot out this time of year, but the difference between 2 cores is what bothers me, i dont remember it being that much

here is an example after playing modern game for a few hours:

as you can see only 1 core is lower, at 91, the rest is almost the same
but at lower stress tests it could be 84 and 69 with difference of 15, for example
and it always the same cores

 

[micbp]

ps. if im doing throttlestop stress test, it goes to something like this all the time:

Spoiler

so, as you can see, 98 to 87, sometimes even bigger difference

 

[Bazzy 505]

There's always some variance among cores, and like jay mentioned, temperature probes can be a bit off. It's not as common occurence, but it happens nonetheless. No in the bigger scheme of things it will not affect longevity of your system in a meaningful way.

To give you an example, i bought an HP625 laptop on which one of it's turion II cores would always report being 8-10 deg hotter than the other. I have had that laptop since 2010 and it's still running fine now. And lemmie tell you the first 5 years it's been running for 10-12 hours daily.

And before anyone starts asking why do i still have it , i not ashamed to admitt, i'm bit of hoarder when it comes to computers ;p

 

[micbp]

well, thanks guys, but i want to highlight 2 points which are either missed or not important (im not a specialist, so i cant be sure)

1. when idle, temperatures are almost exactly the same (+/-1 degree), so i think probes are probably not off? unless they start to go off at a higher temperatures

2. the concern is because i dont remember such difference between cores prior repasting

and it always the same cores, 4th is always hotter, 5th is always cooler than the rest, under stress
with some of you saying its not common, and other saying i should check things and the fact that it wasnt like this before, makes me think that something was actually done wrong

the reason i ask, not because of longterm impact, but rather because it was paid for with quality guarantee (i know it could be done at home as well, but i dont want to mess with something like repasting, since im not really agile this way)

to put it blantly, im deciding if i should call there and tell that something is wrong with the differences in the cores temperatures now, and that they should re-do it

 

[Bazzy 505]

well, thanks guys, but i want to highlight 2 points which are either missed or not important (im not a specialist, so i cant be sure)

1. when idle, temperatures are almost exactly the same (+/-1 degree), so i think probes are probably not off? unless they start to go off at a higher temperatures

2. the concern is because i dont remember such difference between cores prior repasting

and it always the same cores, 4th is always hotter, 5th is always cooler than the rest, under stress
with some of you saying its not common, and other saying i should check things and the fact that it wasnt like this before, makes me think that something was actually done wrong

the reason i ask, not because of longterm impact, but rather because it was paid for with quality guarantee (i know it could be done at home as well, but i dont want to mess with something like repasting, since im not really agile this way)

to put it blantly, im deciding if i should call there and tell that something is wrong with the differences in the cores temperatures now, and that they should re-do it

in that case you should take it back to the shop, chances are they did not re-attached the heatpipe over the CPU die correctly, bent the assembly or did not not apply paste correctly. Out of curiosity. why did you have it repasted ? It's a coffee lake laptop, it's not old enough to be needing such service.

 

[drea.drechsler]

You might also look at a monitoring program (HWInfo64 would be good) and track thread or core utilization across a period of time. You might find that for some odd reason Windows' scheduler may tend to favor one core over another when scheduling processes onto them. I've found that to be the case with my processor and don't understand why.

 

[micbp]

so you guys do think that this is wrong
ok, thank you for replies

@Bazzy 505
well, it says that gaming laptop should be repasted once a year at the least, and this one often runs at 100 degrees during gaming, so im doing a repasting once a year, since paste pretty much turns into powder at this point. am i overdoing it?

@drea.drechsler
thank you, yes, i noticed sometimes it prioritizes other cores also, and they get higher temp (during random desktop tasks, but those temps dont go beyond 70 usually)
but during big stress tests, modern games, its always this one that gets around 5 degrees hotter and the one below usually 7-10 degrees cooler than all the rest
i think throttle stop shows utilization during it stress tests (C0%) ?? it is consistant with all cores (goes to 100 and then drops down simultaneously at all cores as soon as it throttles)

 

[drea.drechsler]

...
but during big stress tests, modern games, its always this one that gets around 5 degrees hotter and the one below usually 7-10 degrees cooler than all the rest
...

And a last thing to consider is there are going to be inconsistencies and defects in the silicon (die) not just between different CPU samples but across the die of an individual sample. We're familiar with what overclockers call a "golden cpu" which is one that overclocks higher or runs cooler and at lower voltages. It seems only logical the same thing can be seen with different cores on the same die. So it should naturally follow that some cores will have higher leakage current and therefore will run hotter even with identical workloads.

AMD recognizes this phenomenon since 3rd gen Ryzen with core ranking and CPPC, a protocol for communicating the core ranking to the OS. Coupled with an OS scheduler that's architecture aware -- so it knows how cores share CPU resources -- it can favor those cores for scheduling threads. I have to believe something similar happens with Intel CPU's since the physics at work are the same. How it's dealt with is probably different, of course.