Tech Support Forum banner
Status
Not open for further replies.

Correct CPU temperature readout?

9K views 61 replies 7 participants last post by  greenbrucelee 
#1 ·
Hi guys,

The value my motherboard and the CPU itself reports on its temperature are very different; which one is more trustworthy?



This is the HWMonitor screenshot, and you can see both values reported

CoreTemp reports on the CPU's values, I guess:



Images taken at little to no load.

I'm running an AMD FX-8150 at the stock 3.6GHz, the Voltage is 1.292V and constant on either idle/load, I'm a bit short on time but at full, full load (prime95 + furmark) the motherboard reports about 71C, but the CPU values are only about 54C or so. The heatsink itself is not that hot when I touch it (compared to my GPU which runs 70C and up easily and it's red hot).

I'm running this inside a decent case (Zalman Z12 Plus) with two exhaust fans near the heatsink, and a default pull fan mounted on the heatsink itself (it's a CoolerMaster Hyper 212 EVO). The room temperature is about 23C or so, gets hotter during the day but so does the PC. The thermal paste has been applied correctly and even reapplied just in case (aftermarket Noctua NT-H1). It just doesn't seem like it should run at 70+ degrees when it's on stock...
 
See less See more
2
#2 ·
Temperature sensors can misreport actual temperatures. In science, we always calibrate thermometers as even the most expensive can be off by up to several degrees. Check the temperatures reported by software applications at idle against what your BIOS reports. If they are the same then you know that the software apps are reporting correctly. However, to confirm if they were accurate you would have to use a calibrated external temperature probe.

You will get better airflow with a positive pressure system. If you can, move at least one fan to the front of the case where it will push cool air in. Also, I'd mount the heatsink fan so that it pushes, rather than pulls, air through the fins.
 
#4 · (Edited)
Thanks for the reply!

Yes, the case comes with 3 fans, the ones in the back and front are pushing air out, and the one in the front (with a dust filter) is pulling air in, like so:




Oh, and yeah the fan on the heatsink is indeed pulling air from ambience and pushing it through the fins, that's what I meant, perhaps I misused the term.

The BIOS reading is the motherboard's readout of 40+ C, but I think that's really high for not even overclocking and using an aftermarket cooler. I know 212s aren't godlike but they should keep the temps down to meet manufacturer's specifications.

And Panther063, you should look at the value HWMonitor reports at the top, which is 44C, compared to what the sensor on the processor itself reports down below and on the CoreTemp picture. In previous versions of HWMonitor the value below would be represented as Core temperatures, and since it's an FX-8xxx there were 8 values.
 
#6 ·
Panther063, you should look at the value HWMonitor reports at the top, which is 44C, compared to what the sensor on the processor itself reports down below and on the CoreTemp picture. In previous versions of HWMonitor the value below would be represented as Core temperatures, and since it's an FX-8xxx there were 8 values.
The temp reported at the top is for the CPU region on the motherboard, the actual reading is listed at the CPU temps.
 
#7 · (Edited)
Oh!

So what is the important value to make sure that it doesn't exceed the manufacturer's spec?

They say AMD FX chips should not exceed 62 or so degrees in order to not degrade, and will start to throttle at 70 degrees. Is this controlled by the CPU or the Motherboard? In other words, will the top or the bottom value matter?
 
#10 · (Edited)
It was taken at idle, the power saving options are turned off.



This is under prime95 for half an hour or so, it will stay the same for however long I run it after that though

The 3.3V reading is incorrect in windows, I tested on multiple PSUs, BIOS values are fine

The fan RPM reading sometimes shows a 0 for a split second, then just goes back to the normal value. The fan is a 600-1600RPM fan so it's spinning alright if I'm looking at the below temperatures, but the 66C reported by the motherboard is worrying. The CPU doesn't throttle or anything though, always stays a 3.6 on all cores.
 
#14 ·
What I'm more concerned about that your CPU temps is that TMPIN2, which at 128 C is way too high if the reading is accurate. Check to see if the motherboard's heatsinks have their pins firmly attached. If you find one that's come loose then the heatsink needs to be remounted with new thermal compound.
 
#15 ·
Do not worry, HWmonitor shows that as 128C, SpeedFan shows it as -128C, BIOS does not show the value at all. I'm fairly sure it's an anomaly. The heatsinks are not nearly as hot, and they have never been removed. The motherboard is from 2012 and I doubt it would survive for more than two years on those boiling temperatures.
 
#16 ·
Correct
TMPIN2 has shown false highs on most HWMonitor screenshots posted.
Your CPU temperatures are fairly low and no reason for concern, you must live in a cool climate. AMD FX-8150 @ 48c max under a stress test is very good.
Even if it were running at the other reported figure of 66C it would give no reason for concern.
 
#17 ·
While I agree the 66C test was a good one, I tested it again under a slightly warmer ambience (and ran furmark to tax the PSU and GPU max and also just create more heat in the case) and for some reason if I do other things while the stress test is in the background, the temps rise much higher. I actually got it to 75+ last night and I believe the CPU started to throttle because the first two cores ran at 4.2Ghz and the rest of them ran at 1.4Ghz, and the temps dropped a bit. Even if the second value was true, I'm fairly sure my motherboard lowered the clock of the CPU automatically.

Perhaps I could try getting a couple of better fans? I'm not entirely sure about this airflow/pressure deal, but the fins on the 212 evo are pretty spread out, so would good airflow fans be a better solution actually? They run at much lower RPM for the same amounts of air being pushed through. I've been looking at what I can find where I live and not too many exciting fans, nothing from NZXT or Scythe, barely anything from CM, Noctua and Corsair...I've managed to find Noctua NF-P12 PWM and Corsair AF/SP120 Quiet/Performance from the better ones, but would this be a waste of money on a budget heastink such as the 212 evo? Should I get something much less expensive like the Arctic Cooling F12s?
 
#18 ·


these are the temps I'm getting if I just browse the internet while the stress test is in the background. For some reason it makes the temps about 10C higher.

The CPU fan is also spinning at a lower RPM than it should, I remember it spinning very near 1600RPM before, maybe like 1585 or so. Now it's spinning much lower at max RPM.
 
#19 · (Edited)
The Hyper 212 Evo may be a budget cooler, but it rates highly, fitting better fans can only improve it.
Also, although your CPU temps are still fine, I'd be concerned about the GPU reaching 95C.
I've used those Noctua fans before on a CPU heatsink, and they are very quiet and produce a decent air flow, they also have a 6 year warranty.
 
#20 ·
The GPU is hot because it's kind of broken, a drunken idiot smashed into my open case and tore the GPU right off the socket, half the pins were stuck on the GPU and half of them were still in the motherboard bent in all directions, the GPU was horribly bent but I straightened it out somewhat and it works perfectly. It's bent in a way that doesn't allow enough airflow to go into the heatsink so that's why it runs very hot. It's been like that for 3 years almost and I'm going to buy a new one soon anyway, so I'm not worried about that.

The noctua fans then? If I put two of them in push-pull, will it improve performance compared to the stock fan I've got on the 212 Evo? The European version has a 600-1600RPM, 24.9 – 66.3 CFM, 0.3 – 1.7mm H2O at least on the spec sheet, the Noctua fans as far as I've been able to find only list the RPM going from 900-1300. Will the static pressure advertised on the NF-P12 PWMs even matter with as open a heatsink as the 212 EVO? I feel like even a performance fan could blow through that easily on lower RPMs.

Also due to laws in my country, the fans usually have a 2 year warranty regardless of what the original manufacturer offers.
 
#21 ·
speedfan is for fan and should not be trusted as it hasn't been accurate for years. As other have said the BIOS is always most accurate, always trust the BIOS over software.

You can buy after market coolers for graphics cards which sit over the top of the stock fans. These can work well but if your card has been bent there may be un seen damage even if it seems to work ok.

I use a noctua cooler and a friend of mine uses the noctua fans and they are very good. You don't really need push-pull as long as it is taking the air of it then it should dramatically lower the temps.
 
#22 · (Edited)
Yes, I am planning on buying a new GPU soon, probably an R9 270x, perhaps the ASUS DirectCUII edition which has good and silent cooling, I'm not going to mod the cooler on the GTX460 since it's an old card and it's not worth it at this point. Whether or not it sustained some unseen physical damage such as the heatsink being bent somewhere I don't know, I did open the card and apply new paste just in case and everything but it hasn't suffered in either stability, benchmarks or game performance. It's only a lot hotter and sounds like a jet now.

So what you are saying is one of the noctua NF-P12s will reduce the temperatures compared to the stock CM cooler which seems to be an XtraFlo 120 (though with reduced RPM for the European version)? I know it will be more silent due to the reduced RPM but once again I wonder if static pressure will matter a lot with this heatsink that has a lot of space for air to flow in unlike something like radiators, the NF-P12 being optimized for that and all. I'm not worried about the noise of the stock fan because it's really inaudible and I don't have the case close to me either so I can barely hear it ramp up the volume.

Do you think the NF-P12 fan or the Corsair AF120 High Performance 1650RPM edition will be the better choice? Perhaps an SP120? They have almost the same prices and I'm not sure which one will benefit me more. If you have no experience with the Corsair fans then I'll just get the Noctua one. Thanks for helping me out in either case, it's much appreciated!
 
#24 ·
Okay, I'm going to go with a Noctua NF-P12.

The only thing is, I can only find a NF-P12 at a reasonable price, without the PWM at the end.

How much will PWM matter? Does the non PWM P12 have problems with noise or something, or will I be okay with the non PWM version?
 
#25 ·
A DC motor can be slowed to a degree by applying lower voltages. A PWM motor uses a square waveform to adjust the speed. Pilots in WWI did something similar when they "blipped" the throttle on their early rotary engines, which operated at either full speed or off. With a PWM fan the time the full voltage is applied in relation to the time it is turned off (i.e., the duty cycle of the waveform) determines the speed.


Duty cycle:



A PWM fan's speed can be changed via software if plugged into a supporting 4-pin motherboard fan header or digital fan controller. However, the speed of non-PWM fans can be changed too with thermistors, in-line resistors, inexpensive potentiometer fan controllers, diodes or even different combinations of wiring. Noctua includes speed reducers with many of their fans.
 
#26 ·
Thanks for that explanation!

My ASUS board has Q-Fan control for 3-pin chassis fans, I've tried it before and it worked with an old CoolerMaster 12cm LED fan. It would vary from around 800 at idle to 1200 (spec RPM) at full load, depending on the CPU, and it's totally adjustable at bios, what temperature and what percentage to use, I'm fairly certain I won't have much trouble with a non-PWM fan.

Also, I can adjust the stock CM cooler to run at similar RPM and try them in push-pull just to see what the performances would be like.
 
#30 ·
4 pin fans are pwm(pulse-width modulation) 3 pin fans are voltage controlled(a positive and negative wire plus a fan speed sense wire) Most retail motherboards will control the voltage on a 3 pin fan some have a bios setting most are auto-detect. Rule of thumb if you have 3 wire chassis fan headers which control the chassis fan speed it'll control a 3 wire cpu fan plugged into the 4 wire header just fine.
 
#32 ·
I've got an update to post here

Getting the NF-P12 was very bad advice since I've installed the Noctua fan but it didn't help the temperatures at all. I'm really disappointed because I was thinking that the noise would be the only thing the NF-P12 would help with, and because the stock fan is silent as it is, it really didn't help that either. The temperatures are exactly the same, maybe even a bit worse. I've put the stock cooler back on.

The motherboard will reduce clock speeds of various CPU cores when it reaches 75C as indicated by the motherboard sensor. They will start working on 1441Mhz instead of ~3600Mhz as I've seen on CPU-Z. Even though it's prime95 + furmark unrealistic loads, I think that on stock with an aftermarket cooler and fan is absurd.
 
Status
Not open for further replies.
You have insufficient privileges to reply here.
Top