cpu thinks it is overheating but itsn't

[mittacc]

I've had this problem for months and I think it's time to find the root of the problem now.

When my computer has been turned off for an extended period of time (like when I sleep) something resets in it that causes the cpu to think that it's about 95-100 degrees celcius, even if not under load. It usually stops being this way after booting a demanding game and playing for a bit and then turning it off. No hot air is coming from the cooler when it says it's overheating (i have a AIO-water cooler) and when it actually is under full load in normal circumstances it has never gone past 55 degrees so these 100 degrees are bloody impossible. While it is like this it throttles my performance as well :c

This is was it looks like: https://gyazo.com/1d508c2cf01b5c836f998bf22a14c4f2

edit: This is how it usually looks while idling https://gyazo.com/2cf16aaf74f789350bf170d9d8957473

[Deleted]

It's unlikely that the sensors inside the CPU are broken so I think your CPU IS getting that hot - have you checked the CPU itself? (try to touching the socket or the mainboard close to the CPU)

It sounds like your cooler is not working sometimes, e.g. an air bubble forming in the pump during the long shut down.

That would explain why the radiator air is really cool since the heat stays at the CPU. Eventually the water will start boiling and that might get the water flowing again. Turning it off and on again could fix that too.

Be thankful that the CPU does throttle down otherwise you'd be needing a new CPU by now

[Evildeffy]

You sure your AIO is in full working condition?

Next time it happens touch the tubing at the CPU block, if that's hot (not luke warm, I mean pretty warm/hot) then your pump is failing.
This could be, dependant on type of AIO, that you're feeding it power with a 3 or 4-pin fan connector and set it to auto/low RPM on that port and it doesn't kick on.
This should be changed into 100% full blast power and if it's fed through SATA/Molex power ... well then you just have a failing pump in general and needs replacing.

There's also the possibility (extremely rare) that the temperature sensor is busted, in which case you're fucked regardless and can't do anything about it.
Whether it's on the motherboard or on the CPU has no relevance, either one would require replacement for it to be "fixed".

Also a possibility is that your IHS' and your CPU die are no longer making proper contact (dried up TIM), you could contact a de-lidding service to de-lid and re-lid your 4670K with some liquid metal TIM since you're out of warranty anyway, I recommend doing this regardless since the temp gains on a 4670K/4770K are quite massive.

Having said that... I've seen a lot of issues with that very CPU and Mobo combo with a friend of mine, including high temps, and it could be a coincidence but always just general instability and annoyances overall, she got so frustrated by it she ditched the thing entirely.. can't even stand to look at the thing anymore.

[mittacc]

Now you're just scaring me :s I haven't tried touching the cpu itself while this is happening but I will test it the next time it happens.

My AIO is the nepton 240M and I don't have the receipt since it was a gift so using a warranty probably isn't an option :/

[Evildeffy]

Now you're just scaring me :s I haven't tried touching the cpu itself while this is happening but I will test it the next time it happens.

My AIO is the nepton 240M and I don't have the receipt since it was a gift so using a warranty probably isn't an option :/

Do note:
I did not say to touch the CPU itself but the waterblock which houses the pump and the 2 tubes coming out of it.
This is placed as a heatsink directly on top of the CPU.. so you're not touching the CPU but the heatsink of the CPU.
And as said if the tubes that are connected to that pump feel warm/hot and colder (noticably) near the radiator then your issue has been identified.

As far as warranty goes, I was referring to the CPU to this one, not the cooler.

Please try to re-read my post slightly slower and no need to imagine anything of the sort for fear, just approach it calmly.
And for your reference when re-reading: IHS = Internal HeatSink, this is referring to the metal plate on the CPU itself.
This is not something you could verify yourself if you're not an experienced builder/tweaker, hence why I said to bring it to a de-lid and re-lid service.

This is done only if other options are exhausted, the basics always first.

[Blupparen]

No hot air from cooler can mean that its not connected well to the cpu. Bad connection or "seating" = bad heat transfer.

[Lathais]

Do note:
I did not say to touch the CPU itself but the waterblock which houses the pump and the 2 tubes coming out of it.
This is placed as a heatsink directly on top of the CPU.. so you're not touching the CPU but the heatsink of the CPU.
And as said if the tubes that are connected to that pump feel warm/hot and colder (noticably) near the radiator then your issue has been identified.

As far as warranty goes, I was referring to the CPU to this one, not the cooler.

Please try to re-read my post slightly slower and no need to imagine anything of the sort for fear, just approach it calmly.
And for your reference when re-reading: IHS = Internal HeatSink, this is referring to the metal plate on the CPU itself.
This is not something you could verify yourself if you're not an experienced builder/tweaker, hence why I said to bring it to a de-lid and re-lid service.

This is done only if other options are exhausted, the basics always first.

Actually, I think he should be just a bit scared. I know if my CPU was reading temps like that and throttling I'd be a bit worried. Instead, he just brushed it off and said, "Nope, it's not really overheating, it's just acting like it is." He should have been scared that it is overheating right from the start.

Now, that's not to say that his CPU is gonna die because of it or anything. It is definitely a solvable issue. But yeah, he should be scared his CPU is overheating, not just brushing it off saying it's not when it clearly is.

[mittacc]

Actually, I think he should be just a bit scared. I know if my CPU was reading temps like that and throttling I'd be a bit worried. Instead, he just brushed it off and said, "Nope, it's not really overheating, it's just acting like it is." He should have been scared that it is overheating right from the start.

Now, that's not to say that his CPU is gonna die because of it or anything. It is definitely a solvable issue. But yeah, he should be scared his CPU is overheating, not just brushing it off saying it's not when it clearly is.

Well, I did shit my pants the first time it happened but it has been happening for about a year and I just haven't been bothered searching for a solution until now

- - - Updated - - -

No hot air from cooler can mean that its not connected well to the cpu. Bad connection or "seating" = bad heat transfer.

True but wouldn't it consistently be overheated if that was the case?

[chazus]

No heat felt from vent fans, which could indicate that AIO cooler is not connected, and heat isn't transferring for one reason or another. If the vent is hot, that means at the very least, it's doing it's job. It's not, if overheating is happening.


Is the system throttling or chugging at all, or does it work great 'except for the fact that monitors say its running hot'?

[Evildeffy]

Actually, I think he should be just a bit scared. I know if my CPU was reading temps like that and throttling I'd be a bit worried. Instead, he just brushed it off and said, "Nope, it's not really overheating, it's just acting like it is." He should have been scared that it is overheating right from the start.

Now, that's not to say that his CPU is gonna die because of it or anything. It is definitely a solvable issue. But yeah, he should be scared his CPU is overheating, not just brushing it off saying it's not when it clearly is.

Of course you don't want this point to happen but my point is that someone who's scared shitless will pay less head to what it is I wrote.

That in turn makes the process a lot more difficult and often not entirely clear because logic and clarity isn't being applied properly.

Hence my statement of not to approach it with fear but calm down and read the post thoroughly and precisely as it will sort a lot out if he sticks to it.

- - - Updated - - -

No hot air from cooler can mean that its not connected well to the cpu. Bad connection or "seating" = bad heat transfer.

You would be right in the case of a normal heatsink as it is just a giant lump of metal doing the transfer with a fan.
However since he has an AIO it introduces another middle-man: The waterpump.

If the pump has failed the contact between the coldplate and CPU IHS can be rock solid but you'd still be fucked regardless.

Hence why I suggested the literal simplest of things to try first as it can eliminate and conclude a lot of issues.

- - - Updated - - -

Well, I did shit my pants the first time it happened but it has been happening for about a year and I just haven't been bothered searching for a solution until now

This kills PCs and being the tech nerd that I am... this hurts my soul and heart to read -.-

True but wouldn't it consistently be overheated if that was the case?

Would fully depend on the load and the actual contact made, if it's a contact issue in the first place.
It's not that your entire CPU is not making contact, but parts of it most of the time.

Thereby having an over-time effect.

[mittacc]

Okay so it happened today again and I decided to lock all the system fans to max speed (everything except the pump and the radiator fans are connected to a fan controller so they aren't affected). The moment I saved and exited the bios the radiator fans, which are set as cpu fans, turned quiet. Hopefully it fixed it.

https://gyazo.com/bb761b456a27f5914ae225e574ba7114

edit: Forgot to say that the tubing next to the pump was a little bit hotter than the rest but not too much

[Deleted]

Okay so it happened today again and I decided to lock all the system fans to max speed (everything except the pump and the radiator fans are connected to a fan controller so they aren't affected). The moment I saved and exited the bios the radiator fans, which are set as cpu fans, turned quiet. Hopefully it fixed it.

https://gyazo.com/bb761b456a27f5914ae225e574ba7114

edit: Forgot to say that the tubing next to the pump was a little bit hotter than the rest but not too much

How long did the system run in the overheated state? If it was only for a short time then the tubing won't be very hot yet.

The system fans (excluding pump and radiator fans) are going to have little to no effect on the CPU temperature as that is (almost) solely dependent on the pump and the radiator.

Also, are you sure the fan controller responsible for the pump and radiator is working correctly? And where does it get its temperature information from?

[Evildeffy]

Okay so it happened today again and I decided to lock all the system fans to max speed (everything except the pump and the radiator fans are connected to a fan controller so they aren't affected). The moment I saved and exited the bios the radiator fans, which are set as cpu fans, turned quiet. Hopefully it fixed it.

https://gyazo.com/bb761b456a27f5914ae225e574ba7114

edit: Forgot to say that the tubing next to the pump was a little bit hotter than the rest but not too much

Next time let it go on for a few minutes and then check tubing temperature.
There should be technically no discernible (by your hands) difference between top and bottom.

Letting it go on for a few minutes exacerbates the issue and if the temperatures become more disparaged between top and bottom ... well then you have your answer.

I'm thinking it is still your pump in this regard.

[mittacc]

The system fans (excluding pump and radiator fans) are going to have little to no effect on the CPU temperature as that is (almost) solely dependent on the pump and the radiator.

The pump fan is the only "fan" connected to the "sys fan" slots while the radiator fans are to the cpu fan slot. The others run through an NZXT GRID+ V2 fan controller and are controlled by software directly from temoerature data readings.

- - - Updated - - -

Next time let it go on for a few minutes and then check tubing temperature.
There should be technically no discernible (by your hands) difference between top and bottom.

Letting it go on for a few minutes exacerbates the issue and if the temperatures become more disparaged between top and bottom ... well then you have your answer.

I'm thinking it is still your pump in this regard.

It ran for about 5 minutes. I did feel a temp difference between the very "base" of the pump and the rest of the tubing before I changed the sys fan speed in the bios.

I think you're right about the pump but it's so daunting for me to switch the thing since I basically have to dismount the entire rig because the cut out in the back of the case is a few mm too small for the pump mount and I'm always shitting bricks while changing stuff in the computer '

[Denpepe]

euhm shouldent the pump be connected to the CPU fan and the fans from the radiator to the ones on the pump? my old corsair cooler worked like that?

[Evildeffy]

euhm shouldent the pump be connected to the CPU fan and the fans from the radiator to the ones on the pump? my old corsair cooler worked like that?

Technically doesn't matter as long as the port is set to no automatic fan regulation.
Even then the software SHOULD override mobo settings and regulate it as well.

[mittacc]

It has yet to happen again after I did that change in the BIOS for 4 consecutive days now...may I call it fixed?

Thanks for the help!

[Evildeffy]

It has yet to happen again after I did that change in the BIOS for 4 consecutive days now...may I call it fixed?

Thanks for the help!

Well you can consider it fixed for yourself but your change has done nothing to warrant the "fix".

So whilst I wouldn't consider it fixed you can if you want but I recommend you to, at all times, keep an eye on temperatures because simply changing case fan settings to full blast should not EVER alter the issue you had if the power was not affected for either pump or radiator fans.

[Deleted]

Well you can consider it fixed for yourself but your change has done nothing to warrant the "fix".

So whilst I wouldn't consider it fixed you can if you want but I recommend you to, at all times, keep an eye on temperatures because simply changing case fan settings to full blast should not EVER alter the issue you had if the power was not affected for either pump or radiator fans.

I think we both misread that. It seems the pump and the radiator that are connected to the main boards fan outlets, while all other fans had their extra controller.

In that case the BIOS not suppling enough power to those outlets can cause the issue of the pump not working and the CPU overheating. It would also explain why the issue eventually fixed itself when the heat inside the case went up far enough.

Locking those outlets to max then can fix the problem, but I'd still keep my eyes open.

[Evildeffy]

I think we both misread that. It seems the pump and the radiator that are connected to the main boards fan outlets, while all other fans had their extra controller.

In that case the BIOS not suppling enough power to those outlets can cause the issue of the pump not working and the CPU overheating. It would also explain why the issue eventually fixed itself when the heat inside the case went up far enough.

Locking those outlets to max then can fix the problem, but I'd still keep my eyes open.

I didn't misread that, he stated that early on that the only fans connected to the mobo are the pump and radiator fans.
But he states directly that he turned up the system fans and not the pump/radiator fans.

His English could indeed be terrible and have explained it wrongly but I'm assuming he didn't.

But yes as the Nepton 240M is connected via 4-pin PWM it should have been self-regulating the pump speed to max as well but it is plausible that if, for whatever reason, the BIOS is suck and put the fan NOT on PWM regulation by default even though it has 4-pin but sticks it on a certain %-age.. then yes it could happen.
It would be a pretty big fking oversight and honestly I do not expect GigaByte to make such errors.

But who knows, let the OP correct us here and if his Engrishu was poor to state that as well