Question Laptop CPU temps fluctuating even at same wattage and fan speed ?

[ellanix]

Laptop model: Lenovo Legion 5 15AC6H

Laptop specs
CPU - Ryzen 7 5800H
GPU - RTX 3060 130w
RAM - 16GB
Storage - 2TB

So, I have been having this issue since the past few months where i would notice that at a power level of say 25 watts, the temps would sometimes be 55 C and sometimes as high as 80 C. This was all at roughly the same room temp and the exact initial fan speed of 2500. what would happen is at idle it would be at 5 to 10 watts and temps at 45 to 50. then after running a load, id get those wild variation of temps.

I have tried repasting it several times. for the first few days, it runs predictably and is cool (85 C at 80w for a laptop). but then after a week, it starts doing these weird things. the reason i am posting this is because when i increase the power limit to 54w ( the cpus actual tdp), itll go to 100 C even though just a week ago it ran 80w fine. if the issue is the contact between the cold plate and die getting wonky, how to solve it?

The gpu temps remain constant and predictable always, so i think its the contact getting off because of some reason or AMD doing weird stuff like if all 25w is going to 1 core, it will get very hot and if the 25w goes to all cores, then overall temps are going to be low. but it still doesnt explain the 100 C at 54w under all core load after just a week of repasting.

 

[Grobe]

Could you provide a CPU load vs. temperature graph, when the cpu tamp regulates normally and when it fails to operate normally ?

 

[ellanix]

Could you provide a CPU load vs. temperature graph, when the cpu tamp regulates normally and when it fails to operate normally ?

how do i do that

 

[beyondlogic]

My laptop specs -
Legion 5 15AC6H
CPU - R7 5800H
GPU - RTX 3060 130w
RAM - 16GB
Storage - 2TB

So, I have been having this issue since the past few months where i would notice that at a power level of say 25 watts, the temps would sometimes be 55 C and sometimes as high as 80 C. This was all at roughly the same room temp and the exact initial fan speed of 2500. what would happen is at idle it would be at 5 to 10 watts and temps at 45 to 50. then after running a load, id get those wild variation of temps. i have tried repasting it several times. for the first few days, it runs predictably and is cool (85 C at 80w for a laptop). but then after a week, it starts doing these weird things. the reason i am posting this is because when i increase the power limit to 54w ( the cpus actual tdp), itll go to 100 C even though just a week ago it ran 80w fine. if the issue is the contact between the cold plate and die getting wonky, how to solve it. the gpu temps remain constant and predictable always, so i think its the contact getting off because of some reason or AMD doing weird stuff like if all 25w is going to 1 core, itll get very hot and if the 25 w goes to all cores, then overall temps are going to be low. but it still doesnt explain the 100 C at 54w under all core load after just a week of repasting

Honestly it's quite common on laptops the issue is heat pipe design of that model shares the same heat pipe over both GPU and CPU you may be able to get some better thermal pads to help dissipation better but what happens is the heatpipe can only take so much heat from one component each of them has 1 dedicated pipe but it's clearly not enough

 

[ellanix]

Honestly it's quite common on laptops the issue is heat pipe design of that model shares the same heat pipe over both GPU and CPU you may be able to get some better thermal pads to help dissipation better but what happens is the heatpipe can only take so much heat from one component each of them has 1 dedicated pipe but it's clearly not enough

sorry for creating a misunderstanding. the CPU temp variations are irrespective of gpu load. the gpu load was always at around 15w. so there was no impact of varied load on the heat pipes

 

[beyondlogic]

sorry for creating a misunderstanding. the CPU temp variations are irrespective of gpu load. the gpu load was always at around 15w. so there was no impact of varied load on the heat pipes

Double check the thermal pad position make sure there are no thermal pads ripped or damaged again heatpipe design is still over a GPU and CPU that's still going to effect it. Regardless of load the extra wattage may be being dumped to GPU it can also be that a thermal pad isn't making proper contact as thermal pads sometimes act as spacers you can mitigate this by using a thinner thermal pad then the current ones e.g if it's a 2.0 you can go to 1.5 this brings the die closer but you should be wary of going to tight or you can crack the die.

Also when tightening you should be doing a little between each screw at a time if you tighten to much on one side you can bend or warp the heatpipes

 

[ellanix]

Also when tightening you should be doing a little between each screw at a time if you tighten to much on one side you can bend or warp the heatpipes

yes ive done that. i learnt about it gamers nexus teardown videos of GPUs.

Double check the thermal pad position make sure there are no thermal pads ripped or damaged again heatpipe design is still over a GPU and CPU that's still going to effect it

i think you are talking about thermal paste for CPU contacting the cold plate, as i dont use thermal pads. as for VRM thermal pads and VRAM thermal pads, i think i need to check them. yea, that could be a reason. i will check them

 

[beyondlogic]

yes ive done that. i learnt about it gamers nexus teardown videos of GPUs.


i think you are talking about thermal paste for CPU contacting the cold plate, as i dont use thermal pads. as for VRM thermal pads and VRAM thermal pads, i think i need to check them. yea, that could be a reason. i will check them

Nope talking about the thermal pads around the GPU and CPU area if there not in the correct places it won't make proper contact. This shows the underside of the cooler:

https://www.aliexpress.com/item/100...782df7146c99161f62327734acd&afSmartRedirect=y

 

[ellanix]

Nope talking about the thermal pads around the GPU and CPU area if there not in the correct places it won't make proper contact. This shows the underside of the cooler:

https://www.aliexpress.com/item/100...782df7146c99161f62327734acd&afSmartRedirect=y

i checked it and i dont have the thermal pads which are there on top of the dies. i also dont have the thermal pads contacting the heat pipe. but as far as i remember, i never had them in the first place. ill buy them and see if it works

 

[beyondlogic]

i checked it and i dont have the thermal pads which are there on top of the dies. i also dont have the thermal pads contacting the heat pipe. but as far as i remember, i never had them in the first place. ill buy them and see if it works

That would explain it possibly the pads where not added properly In factory

 

[ellanix]

ok so i think i know why it is happening. i think 2 days ago again i saw it get to 80 c for 25 watts. then i remembered that a few hours before that i was holding the laptop with hand by the corner of it. i think that since the chassis of the laptop is all plastic and the fact that the cooper cooler system which is metal is directly attached to the chassis, i think that because of holding the laptop on one side, it put a little pressure on the cooling system and it moved a little bit, so now it is not contacting the die properly. the reason i feel this could be the problem is because the laptop is actually pretty heavy at 2.4 kg. holding an object of that weight which is made of plastic from just one side is sure to affect the chassis. so now i guess ill have to repaste it and not hold it like that and see what happens

 

[ellanix]

i experimented with flexing the chassis while it was under load, and yes, the temps were fluctuating. the chassis of this laptop is not sturdy enough to hold it in one hand. itll start to affect the contact of the coldplate and die. so yea, best thing i can now do is to just always handle it by both hands.