Question Deciphering if temps are false or not

[white.a.drew]

Is there any way to find out if a motherboard is sending flase readings to hwmonitor, speccy, cpuid, and etc

For a while now I've been noticing a odd issue mainly between two programs on older systems..... Not old old but it's an older system I am facing the issue with now. However this system might not be giving false readings I can't be sure on this though because like I said I have had issues where the system just isn't providing proper readings and other times where everything is fine....


Is there any way to figure out if temp readings are right or if the sensor is shot/not reading right one. Or is it basically ride or die and just risk the temps being false...

The main reason I'm thinking their false temp readings is, the moment the PC fully boots I open 2 apps the monitor the temps hwmonitor and speccy, they both right off the bat say 115c for the mainboard temp and the other one will say 112c mainboard temp.... But neither ever agree on the temp the mobo is they just right away say ove 100c

 

[drea.drechsler]

The best way is to put a thermocouple on the device where you are measuring temperatures: like the FET's, or CPU heat spreader. But that's problematic because it's usually hard for most people to place one underneath a heatsink. The easiest way is to use an IR thermometer in the close vicinity of the part in concern. Obviously neither will perfectly match your on-board sensors because neither are measuring at the exact same spot. All you'd be looking for is whether they track in sync with changes and have a reasonable difference.

But then, any on-board sensor isn't very accurate either since they're not calibrated. All they are really good for is a general indication of the thermal state so shouldn't be relied on if precise temperature readings are required.

What are your system specs? Ryzen CPU's in particular may seem strange to new users because they are so dynamic. They also have a lot of sensors buried right in the silicon so what you're seeing is the hottest spot at that exact moment, changing from moment to moment as it updates. That leads to rapid fluctuations in temperatures being reported as it boosts and then moves processes around to different cores for wear leveling.

 

[Karadjgne]

There's multiple places on a motherboard that can and do read in the 100°C+ range, at any given moment, depending on use. Sata controller, super i/o, pch etc.

Everything accessible has an address, it's how windows knows where to send stuff and retrieve stuff from, so windows doesn't ask the hdd to send a picture file to ram and then the cpu to be put on screen, it says ide123:003 send xxx.jpg to ifexx123:001 via rmhrty:0432 etc.

Software is written by ppl. They'll put their best guess as to what addresses certain items generally are located at, like the old Com1 serial port was i/o port 0x3F8 irq4, but there's no guarantee that will be the same between vendors, or the same between users and their equipment. Same applies to thermocouples used on the motherboard. Speccy might be reading the Super I/O, Hwmonitor might read the Sata controller, depending on what the software says those particular addresses should be.

I used to use SpeedFan a lot, as it was basically the only fan controller software that was tailorable and I wanted the case fans to respond to gpu temps. 2 of the temps were definitely wrong. One read 255°C constantly and the other read -125°C. Both physically impossible. Meaning those particular addresses were not actually thermocouples, but something entirely different and the voltages present were being interpreted incorrectly.

Just as Hwmonitor has tmpin1-6 that could be anything, totally unknown addresses or results.

 

[white.a.drew]

There's multiple places on a motherboard that can and do read in the 100°C+ range, at any given moment, depending on use. Sata controller, super i/o, pch etc.

Everything accessible has an address, it's how windows knows where to send stuff and retrieve stuff from, so windows doesn't ask the hdd to send a picture file to ram and then the cpu to be put on screen, it says ide123:003 send xxx.jpg to ifexx123:001 via rmhrty:0432 etc.

Software is written by ppl. They'll put their best guess as to what addresses certain items generally are located at, like the old Com1 serial port was i/o port 0x3F8 irq4, but there's no guarantee that will be the same between vendors, or the same between users and their equipment. Same applies to thermocouples used on the motherboard. Speccy might be reading the Super I/O, Hwmonitor might read the Sata controller, depending on what the software says those particular addresses should be.

I used to use SpeedFan a lot, as it was basically the only fan controller software that was tailorable and I wanted the case fans to respond to gpu temps. 2 of the temps were definitely wrong. One read 255°C constantly and the other read -125°C. Both physically impossible. Meaning those particular addresses were not actually thermocouples, but something entirely different and the voltages present were being interpreted incorrectly.

Just as Hwmonitor has tmpin1-6 that could be anything, totally unknown addresses or results.

So it could just be that the motherboard isn't supported properly in either of the monitoring apps "in a dumbed since of what your saying"?

The best way is to put a thermocouple on the device where you are measuring temperatures: like the FET's, or CPU heat spreader. But that's problematic because it's usually hard for most people to place one underneath a heatsink. The easiest way is to use an IR thermometer in the close vicinity of the part in concern. Obviously neither will perfectly match your on-board sensors because neither are measuring at the exact same spot. All you'd be looking for is whether they track in sync with changes and have a reasonable difference.

But then, any on-board sensor isn't very accurate either since they're not calibrated. All they are really good for is a general indication of the thermal state so shouldn't be relied on if precise temperature readings are required.

What are your system specs? Ryzen CPU's in particular may seem strange to new users because they are so dynamic. They also have a lot of sensors buried right in the silicon so what you're seeing is the hottest spot at that exact moment, changing from moment to moment as it updates. That leads to rapid fluctuations in temperatures being reported as it boosts and then moves processes around to different cores for wear leveling.

The specs of the system I'm aware of at the moment

There is one exhaust fan one intake fan
Thermal take 850w psu
1tb hdd boot drive and storage only thing in it right now
I5 4670k liquid cooled and at stalk settings right now
Asus z87-k
16gb ram 1600 2x8gb

Edit oh I forgot there's also a gtx 770 gold

 

[drea.drechsler]

.......
I5 4670k liquid cooled and at stalk settings right now
......

Well that's not going to have Ryzen temp quirks...nor AMD's Bulldozer/Piledriver/Excavator era processors that had equally strange quirks.

But adding to the list of reasons for weird temp readings...most of the Super I/O chips that are used for monitoring the temperature sensors on motherboards have several more "channels", if you will, than they have temp sensors on the motherboard. Those unused channels are often left unterminated and so will give random readings to monitoring programs and some are terminated (resistor, capacitor, short, whatever's required) to force an erroneous value so that it's obvious it's not a valid temperature.

One of the best monitoring programs I know of is HWInfo64. The author makes a sincere effort to provide info on each sensor and most popular boards get their actually connected sensors properly labeled. If you find one that's not correct or behaving strangely you can send him a message (an active forum on his web site) with a dump. He usually makes an effort to fix it if he can, or explain it if he can't fix it. He's constantly releasing beta's with fixes people help him with, as well as updates for new boards.

 

[white.a.drew]

Well that's not going to have Ryzen temp quirks...nor AMD's Bulldozer/Piledriver/Excavator era processors that had equally strange quirks.

But adding to the list of reasons for weird temp readings readings...most of the Super I/O chips that are used for monitoring the temperature sensors on motherboards have several more "channels", if you will, than they have temp sensors on the motherboard. Those unused channels are often left unterminated and so will give random readings to monitoring programs.

One of the best monitoring programs I know of is HWInfo64. The author makes a sincere effort to provide info on each sensor and most popular boards get their actually connected sensors properly labeled. If you find one that's not correct or behaving strangely you can send him a message (an active forum on his web site) with a dump. He usually makes an effort to fix it if he can, or explain it if he can't fix it. He's constantly releasing beta's with fixes people help him with, as well as updates for new boards.

Thank you for your info that helps a lot

 

[white.a.drew]

Well that's not going to have Ryzen temp quirks...nor AMD's Bulldozer/Piledriver/Excavator era processors that had equally strange quirks.

But adding to the list of reasons for weird temp readings...most of the Super I/O chips that are used for monitoring the temperature sensors on motherboards have several more "channels", if you will, than they have temp sensors on the motherboard. Those unused channels are often left unterminated and so will give random readings to monitoring programs and some are terminated (resistor, capacitor, short, whatever's required) to force an erroneous value so that it's obvious it's not a valid temperature.

One of the best monitoring programs I know of is HWInfo64. The author makes a sincere effort to provide info on each sensor and most popular boards get their actually connected sensors properly labeled. If you find one that's not correct or behaving strangely you can send him a message (an active forum on his web site) with a dump. He usually makes an effort to fix it if he can, or explain it if he can't fix it. He's constantly releasing beta's with fixes people help him with, as well as updates for new boards.

this is what i get using hwinfo64

 

[drea.drechsler]

this is what i get using hwinfo64

Not hard to guess your concern is the "temp 2, temp 3, temp 4, temp 5". Since they're unlabeled he either doesn't know what they are, or they're unconnected channels on the super I/O chip. Since they're a nice, stable 100+C even when the system is idle I'd suspect they're open channels, and the next time the system starts up they may even be wildly different. Probing around in the motherboard looking for components running that hot while idle with an IR thermometer might set your mind at ease.

The CPU (weighted value) is very perplexing and I don't understand what the CPU (PECI) values mean; possibly the CPU internal temp sensor? The other CPU value is suspiciously low, 28-33C, might be a socket sensor that's typically lower than CPU internal sensors are reporting.

Probably worth a posting in his forum to see if anyone else knows more about this, or maybe someone here will chip in.

EDIT: found this on PECI....so it looks like that's the only true temperature of the CPU internals.