CPU overheating under low load?

[phanbuey]

Is the cooler hot when this happens, or does it feel cool to the touch. Sounds like something warped and isn't making good contact... or the thermal diode is going nuts.

 

[MacchiatoLatte]

Is the cooler hot when this happens, or does it feel cool to the touch. Sounds like something warped and isn't making good contact... or the thermal diode is going nuts.

It's pretty hot, or at least the air is.

 

[eidairaman1]

Have you physically checked to see if the heatsink itself is getting hot? If the heatsink is getting nearly as hot as the processor, you should be able to feel the heat coming from it without touching it. You could try touching the metal parts of the heatsink too, just be cautious not to give yourself a heatsink-shaped burn.

If the heatsink is hot, the fan is spinning, and it's not chock full of dust, then I'd say check your CPU voltages with CPU-Z, AMD Overdrive, or HWMonitor/OpenHardwareMonitor just to make sure it's not accidentally running it at 1.8v or something.

If the heatsink is cold and you have fresh thermal paste, have you tried reseating the heatsink? I'd like to make another vote for an inexpensive aftermarket heatsink but I don't think that's the problem. I've always changed mine out for aftermarket but that was mostly for looks as well as performance. The stock one should obviously perform well enough to keep it from overheating.

Do thermals change if you power it up on its side? Kind of reminds me of an issue I had back in 2010. I had a Radeon HD 4850 that would overheat unless I laid my PC down on its side (with the card facing straight up and down). I chalked it up to a failed heatpipe and replaced the cooler.


Shouldn't be much. Pulling the battery will actually power everything down and reset rather than just stamping default settings on top of something that may have a glitch or quirk. Sort of like a full, cold shutdown rather than a warm reboot.

Edit: Tons of revisions adding and changing stuff around.

Depends on motherboard actually.

Another thing, make sure there is no plastic film on heatsink.

It's a Corsair CX750m.
Sorry for stupid question by the way, but what's the difference between factory resetting BIOS and clearing CMOS?

How old? New or Used?

 

[moproblems99]

Yes, actually. With no load except HWMonitor, it can stay down near 20-30°.

What I meant is the temperature of the room that the PC is in. Has it gotten warmer or colder since October?

Have you put any other heat generating items near the intakes of the PC? Have you put anything around the PC that could restrict the intake or exhaust?

I don't really believe this has anything to do with the PSU. That wouldn't cause the CPU temps to rise 10c but it could cause the shutdowns. Still, not where I would spend time at this point.

 

[Athlonite]

if the fan is supposed to spin upto 4700+ RPM and now it's not sounds like a dying fan struggling to do its job try replacing the fan if that doesn't fix the problem then there's something more amiss. Have you had any power outages or brownouts or power surges recently those can also cause physical damage to electronic components

 

[xvi]

or the thermal diode is going nuts.

I'm starting to wonder this too. Maybe the temperature sensor on the CPU is failing.
MacchiatoLatte, when you physically check, it sounds like everything seems warm but not hot? Like a fresh brewed cup of tea/coffee hot?

 

[Durvelle27]

What software are you using to read temps

 

[MacchiatoLatte]

How old? New or Used?

It's, I think 4.5 years old? It was new when I bought it.

What I meant is the temperature of the room that the PC is in. Has it gotten warmer or colder since October?

It's gotten colder since October. Maybe 5 degrees outside, 17 or so in my room.

Have you put any other heat generating items near the intakes of the PC? Have you put anything around the PC that could restrict the intake or exhaust?

I moved my fan there, but it's been there for almost a year now.

Have you had any power outages or brownouts or power surges recently those can also cause physical damage to electronic components

Not since December, no.

MacchiatoLatte, when you physically check, it sounds like everything seems warm but not hot? Like a fresh brewed cup of tea/coffee hot?

It's like a slightly-too-hot shower. I'd say around 45 degrees.

What software are you using to read temps

I'm using HWMonitor to read the temps.

 

[JackCarver]

Did you reset BIOS already? If so I assume it's a broken temp sensor which reads cpu temps. You

- applied new thermal paste
- reseated the heatsink
- double checked that heatsink is sitting propperly
- cleaned PC, heatsink, fans
- checked that all fans are working as they should be
- checked that ambient temp is around 17C, would be way too cold for me, but is good for the hardware

So criterion for exclusion says, that it shouldn't show that high temps even at low load. Only thing which is remaining is a broken cpu temp sensor. Maybe time to buy new cpu.

Edit:
By the way, were those cpus soldered or did they use thermal paste like Intel did long time?

 

[MacchiatoLatte]

Did you reset BIOS already?

By the way, where those cpus soldered or did they use thermal paste like Intel did long time?

I did reset BIOS.

I think they're soldered.

 

[JackCarver]

Ok so even on factory setting still heat problems. Broken temp sensor which gives you wrong cpu temps is my guess then. When they are soldered there should also be no problem with the heat dissipation between Die and IHS.

Edit:

I'm using HWMonitor to read the temps.

Did you also check with another program like CoreTemp?

 

[MacchiatoLatte]

Ok so even on factory setting still heat problems. Broken temp sensor which gives you wrong cpu temps is my guess then. When they are soldered there should also be no problem with the heat dissipation between Die and IHS.

Edit:

Did you also check with another program like CoreTemp?

I just checked with CoreTemp and it shows the same numbers.

 

[eidairaman1]

Swap motherboards, test in a shop and see if anything changes...

 

[MacchiatoLatte]

Swap motherboards, test in a shop and see if anything changes...

I don't have a motherboard to swap with, and no shops nearby with hardware, unfortunately.

 

[JackCarver]

I don't have a motherboard to swap with, and no shops nearby with hardware, unfortunately.

You will have to test it with other mainboard, otherwise I assume you won't find the problem.

 

[milewski1015]

It didn't overheat at all. To be exact, it still got to low 70s, but the CPU fan would kick in at that point and keep it from rising more.

By this do you mean the fan isn't starting until the CPU hits the low 70s?

 

[MacchiatoLatte]

You will have to test it with other mainboard, otherwise I assume you won't find the problem.

Shit, that sucks. Guess I'll have to figure something out.

By this do you mean the fan isn't starting until the CPU hits the low 70s?

The fans are always running, but they don't speed up at high temps anymore.

 

[xvi]

Swap motherboards

test it with other mainboard

It would probably be cheaper to buy a low-end CPU used to see if thermals show good there. I would suspect a motherboard would be more costly to ship than a cheap CPU.

If you don't mind my curiosity, Macchiato, are you in the US? I may have some spare parts.

 

[Aquinus]

The fans are always running, but they don't speed up at high temps anymore.

Maybe the fan on the cooler is beginning to fail. It might be trying to run at full tilt already because there's not enough airflow because by the time the CPU gets that hot, it's already feeding 100% PWM to the cooler. I'd turn the machine off and feel the fan. Does it have a lot of movement up and down? Is it easy to spin without friction slowing it down almost immediately? As I understand it, the stock cooler should be spinning up as high as 4.5k RPMs. If it's considerably lower than that at full tilt, the fan might be on the way out.

With that said, a new cooler is a relatively cheap part. It might be a good place to start.

Edit:

HWMonitor says my CPU fan is spinning around 3400RPM at 100%, but I've seen it up at 5000RPM in the past. It doesn't go that high anymore, though.

I missed this post. I'm even more convinced that the fan might be failing. I'd replace the cooler.

 

[MacchiatoLatte]

It would probably be cheaper to buy a low-end CPU used to see if thermals show good there. I would suspect a motherboard would be more costly to ship than a cheap CPU.

If you don't mind my curiosity, Macchiato, are you in the US? I may have some spare parts.

That's a good idea, I'll probably try that out.

No, I'm unfortunately in Sweden, so my options are fairly limited for new and used parts.

I missed this post. I'm even more convinced that the fan might be failing. I'd replace the cooler.

I'll probably try to do that sometime next week if I find a good replacement. Considering a Hyper 212 Evo, do you know if that's any good?

 

[xvi]

Considering a Hyper 212 Evo

They're usually pretty highly recommended as having good performance while being inexpensive. They're getting a little old now, but it would be a good upgrade over the stock cooler.

If the issue is a failing temperature sensor, a colder CPU should help keep it from shutting down, but I suspect you'll still have some issues under load.

If you're comfortable with it, a free and quick solution that would have a similar effect would be to lower CPU voltage to reduce temps. You would gain colder temperatures but it may make your processor unstable. If you accidentally turn voltage up, it could physically damage the processor or motherboard.

The process would be to find some kind of overclocking software (probably AMD Overdrive which I've just noticed has been discontinued, so here's a mirror) and lower the "CPU VID" voltage which will significantly reduce temperatures. You gain colder temps but the tradeoff is you reduce the stability of your processor. As long as you don't accidentally change your voltage up at all (shouldn't be over ~1.4v when you start out), there shouldn't be any risk.
You'll want to save and close out of anything you're working on and slowly lower CPU voltage down until either the temperatures look good or your computer crashes. If it does crash, when it boots back up, set voltage to slightly above where it crashed.

If that works, you could try both the new CPU cooler and lowered voltages and you should hopefully be back to pretty stable conditions.

 

[MacchiatoLatte]

They're usually pretty highly recommended as having good performance while being inexpensive. They're getting a little old now, but it would be a good upgrade over the stock cooler.

If the issue is a failing temperature sensor, a colder CPU should help keep it from shutting down, but I suspect you'll still have some issues under load.

If you're comfortable with it, a free and quick solution that would have a similar effect would be to lower CPU voltage to reduce temps. You would gain colder temperatures but it may make your processor unstable. If you accidentally turn voltage up, it could physically damage the processor or motherboard.

The process would be to find some kind of overclocking software (probably AMD Overdrive which I've just noticed has been discontinued, so here's a mirror) and lower the "CPU VID" voltage which will significantly reduce temperatures. You gain colder temps but the tradeoff is you reduce the stability of your processor. As long as you don't accidentally change your voltage up at all (shouldn't be over ~1.4v when you start out), there shouldn't be any risk.
You'll want to save and close out of anything you're working on and slowly lower CPU voltage down until either the temperatures look good or your computer crashes. If it does crash, when it boots back up, set voltage to slightly above where it crashed.

If that works, you could try both the new CPU cooler and lowered voltages and you should hopefully be back to pretty stable conditions.

Definitely trying this tomorrow! Do you have a recommended software or is a Google search enough?

 

[lexluthermiester]

@MacchiatoLatte
Welcome to the forums!

Looks like a bunch of good people have already chimed in with input and you could be having one of several different problems.

My recommendation would be to remove the heat sink and fan to clean them out, then reseat the unit and see if cleaning helped. Next, @xvi suggestion of lowering the voltage is good idea to try if your CPU has developed a fault. However, to keep this stable, you should also reduce the clock speed by a double percentage, rounded down to the nearest 5%. For example, if you drop the voltage from the 1.345v you stated earlier to 1.3(a good setting to try) which is about 4% drop, lower the clock speed by 10%. This would be a change from 4000mhz to 3600mhz. Then reboot and watch your temps. If Windows crashes, you'll need to drop your clock speed lower to say 3400mhz and try again. If windows is stable but your CPU still overheats then you very likely have a fault in your CPU, the motherboard or perhaps both, which is unfortunate.

Try this out and keep us posted as to the results.

PS, At this time, it might be wise to consider an upgrade. You can get a Ryzen 5 1600 or Ryzen 7 1700 for very cheap and a matching motherboard and ram for very reasonable prices. That's only a thought, but lets work the problem first.

 

[MacchiatoLatte]

@MacchiatoLatte
Welcome to the forums!

Looks like a bunch of good people have already chimed in with input and you could be having one of several different problems.

My recommendation would be to remove the heat sink and fan to clean them out, then reseat the unit and see if cleaning helped. Next, @xvi suggestion of lowering the voltage is good idea to try if your CPU has developed a fault. However, to keep this stable, you should also reduce the clock speed by a double percentage, rounded down to the nearest 5%. For example, if you drop the voltage from the 1.345v you stated earlier to 1.3(a good setting to try) which is about 4% drop, lower the clock speed by 10%. This would be a change from 4000mhz to 3600mhz. Then reboot and watch your temps. If Windows crashes, you'll need to drop your clock speed lower to say 3400mhz and try again. If windows is stable but your CPU still overheats then you very likely have a fault in your CPU, the motherboard or perhaps both, which is unfortunate.

Try this out and keep us posted as to the results.

PS, At this time, it might be wise to consider an upgrade. You can get a Ryzen 5 1600 or Ryzen 7 1700 for very cheap and a matching motherboard and ram for very reasonable prices. That's only a thought, but lets work the problem first.

I'm gonna try cleaning the heatsink and fan, but it'll have to wait until tomorrow because I don't want to wake family by messing around too much.

As for the clock speed, do I do that with the same software as the voltage?

 

[lexluthermiester]

As for the clock speed, do I do that with the same software as the voltage?

You would be doing that from within the bios. Have you done such before? It's ok if not. There's a first time for everything and we can help walk you through that too.