Weird Thermal Throttling

[Basil12]

Hello everyone,
I recently bought a ROG G703VI and I'm having some trouble with performance deficiency issues.
While on idle, 7820HK CPU seems to be working at about 50 degrees celcius but the strange thing is it jumps to 80 degrees for just a second and then falls down to its normal state as nothing happened. My initial guess was that it was just caused by a glitch, but gaming and some monitoring shows otherwise. While gaming, from time to time I notice a fps loss which lasts about 1 or 2 seconds. I checked the CPU usage and temps with XTU and thermal throttling was spiking like crazy. I tried undervolting about 0.005-0.01V but the problem still persists. I maxed out the fan speed to observe how CPU temp will react. Apart from a few degree decrease, it still hits 97-98 celcius randomly. I've attached the weird screenshots I took from XTU while gaming. Could it be because of a faulty BIOS? Or is it a hardware problem?

[ChazTec]

I think your just going to have to deal with it as the laptop sounds like it doesnt have a beefy cooler. You could try changing the thermal paste like another user said but i dont think thats going to help much especially if its a new laptop. 

[Mnky313]

5 minutes ago, ChazTec said:

I think your just going to have to deal with it as the laptop sounds like it doesnt have a beefy cooler. You could try changing the thermal paste like another user said but i dont think thats going to help much especially if its a new laptop. 

 

14 minutes ago, valdyrgramr said:

Are you using the stock paste or?

I agree, changing laptop thermal paste for higher end stuff like IC Diamond, NT-H1, or similar can sometimes drastically drop temps.

[Basil12]

16 minutes ago, valdyrgramr said:

Are you using the stock paste or?

I do. I did not interfere with the inside as I'm afraid I could make the situation worse. Also it could void warranty

 

8 minutes ago, ChazTec said:

I think your just going to have to deal with it as the laptop sounds like it doesnt have a beefy cooler. You could try changing the thermal paste like another user said but i dont think thats going to help much especially if its a new laptop. 

Cooler sounds quite powerful actually. The problem is, many other people that have the same PC seem to have lower cpu temps. I'm thinking of dusty fans, but wouldn't it also affect the GPU temps too? For now, GPU poses no overheating.

[ChazTec]

9 minutes ago, Basil12 said:

I do. I did not interfere with the inside as I'm afraid I could make the situation worse. Also it could void warranty

 

Cooler sounds quite powerful actually. The problem is, many other people that have the same PC seem to have lower cpu temps. I'm thinking of dusty fans, but wouldn't it also affect the GPU temps too? For now, GPU poses no overheating.

Well if you were playing low end games then you probably wouldnt realise the gpu overheating (same with the cpu) Have you checked the temps on the gpu? If thats overheating then you could try down clocking but this seems like a very bizarre problem. 

[markr54632]

1 minute ago, valdyrgramr said:

Not always, but changing the paste with something like kryonaut, which is not conductive, won't damage anything.  If you're too afraid to do it yourself you could just use the warranty to cover this.  But, if you're in the US I don't think they can enforce that either.

They absolutely can. Changing the paste can cause damage later on down the lane. Not to mention if they dont want to cover it all they have to do is say that you damaged it.

 

If it is under warranty or return policy use it.

[markr54632]

Just now, valdyrgramr said:

How is a proper application of kryonaut going to damage it?

How are you going to prove you applied it properly?

[Basil12]

13 minutes ago, ChazTec said:

Well if you were playing low end games then you probably wouldnt realise the gpu overheating (same with the cpu) Have you checked the temps on the gpu? If thats overheating then you could try down clocking but this seems like a very bizarre problem. 

They are mostly decent games such as Overwatch, Kingdome Come or Just Cause 4. At max load, it reaches 75 or sometimes even 80, but no more than that. Another odd thing is, as you can see in images, it often thermal throttles even if does not get too hot (which is 80-ish)

[Basil12]

Could it also be a faulty thermometer? (or whatever it is called in CPU)

[markr54632]

23 minutes ago, valdyrgramr said:

Take pictures or make a video?  If it overheats it will throttle to protect itself, and kryonaut and many other pastes are not conductive.  So, it would take a lot more than that to damage it "down the lane" if the temps are fixed by changing the paste.

The problem is not with what is right or wrong, but rather what you can prove. To fight the manafacturer's decision would take more money than the laptop is worth. Ask the local representative for your region and your brand if it voids the warranty, honestly for all practical purposes it is enforceable.

 

To the op even if the temp sensor was faulty it is not something you could easily repair.

[markr54632]

Just now, valdyrgramr said:

Um, in the US they cannot enforce you opening it and removing the stickers.  OP already has documented that the whatever the maker did is messing it up.  The maker would have a lot trouble proving that op damaged it by putting on kryonaut in the first place.  Since ya know kryonaut itself cannot damage anything.  Plus, with further documentation, the maker would likely cave anyways.

Legally and ethically you would be right. In the real world it would take too much time and money to fight making it enforceable.

 

 

While kryonaut in and of itself would not cause issues all the manafacturer has to do is say kryonaut is not approved for use in your product, you applied it incorrectly, or you didnt torque the cooler down properly and it gets deemed user damage.

[markr54632]

Just now, valdyrgramr said:

It's easy to prove the crap that the maker used is causing issues due to it already throttling.  Secondly, if they document it they maker likely won't fight it unless they're that stupid.  Too much time and money?  It takes like a quick email exchange.

You obviously want to live in a perfect reality where things go swimmingly or don't have very much experience dealing with warranty centers.

[markr54632]

1 minute ago, valdyrgramr said:

I've dealt with warranty centers a lot, actually.  I get you have some paranoia about them, but in my experience, they usually don't care as much you think.  Now, they might care if you try to do a reflow or something like that.

I have had the exact opposite experience.

 

I will also add it is very easy to damage a bare die by tightening the cooler unevenly or too tight.

[markr54632]

Just now, valdyrgramr said:

Well, that would be user error right there.  That's not their fault at all, and I would side with the warranty center if the user did that.  

The problem is as the consumer you cant prove that you didnt once it is sent in.

[markr54632]

3 minutes ago, valdyrgramr said:

Actually, you can it's called documentation.  Which we already went over.

Even if you do it means nothing.

 

Not to mention kryonaut isnt certified for use in a laptop. It's the equivalent of using non dexos oil in a GM that "requires" dexos oil.

[Basil12]

Guys we are going off topic. Anymore speculation on what the problem could be? I will try changing thermal paste via service, but I'm still timid as they might as well screw up too as I heard a lot from people around here.

[markr54632]

31 minutes ago, valdyrgramr said:

Actually, it does because you have proof now.  When you shove proof in the makers face they tend to cave easily.

Nor is that crap most OEMs use hence the high temps.

It absolutely is because the oem is the one that gets to make that determination.

[Basil12]

Up

[X_X]

On 2/6/2019 at 4:23 AM, Basil12 said:

 I tried undervolting about 0.005-0.01V but the problem still persists.

This isn't going to do much, you need 10 times that or more and adjust cache by the same amount.

 

Check all cores for temperature to see if one core is much higher than the rest. Package temperature tells the highest temperature in the package but not what is generating it. You could run RealTemp or HWiNFO while running a constant load such as FireStarter. Keep your clocks low enough so you are not thermal or power throttling.

 

Sounds like a TIM problem, perhaps best course of action might be to return it as not fit for use.