Overclock.net banner
1 - 11 of 11 Posts

·
Registered
Joined
·
12 Posts
Discussion Starter · #1 ·
headscratch.gif


I've got this strange temperature reading that I can't find any information on. I thought it was a HWinfo glitch, or it was just normal for this CPU until I saw someone posting their HWinfo screenshot. The temperature on their screenshot looks like all the other numbers 30 to 50 Celsius.

link to pic i'm referring to

http://www.overclock.net/t/1487479/am-i-doomed-to-ever-overclock-my-4670k

My readings

http://cdn.overclock.net/6/69/900x900px-LL-69218ad3_WTH.jpeg

EDIT
I was sleepy when I posted this, lets give ya more info.

CPU i5-4570k
not overclocked or overvolt I've disabled that because of the stock cooler.

ASRock Z97 Extreme 4

The sensor in question is the Nuvoton NCT6791D super IO Chip shows a steady 74 to 76 Celsius
REALTEMP does not even have this setting.
ASRock A-tuning also does not have a nuvoton sencor.
So the real question is..
Is this a HWinfo glitch or is this a accurate reading and I might have a bad Motherboard. Has anyone else seen this on a ASRock Z97 Extreme 4 ?
 

·
Premium Member
Joined
·
9,160 Posts
HWInfo is just reading hardware. It detects sensor chips on the board (actual hardware chips) and prints their output. While it can theoretically have bugs, in all likelihood 99% of the issues are related to the hardware of the board or also the BIOS.

Notice for example that you didn't read the CPU hardware portion there but another sensor from another company (Nuvoton.. something).

Those chips appear to have multiple inputs around the board, e.g. it acts like a regular multimeter reading the VIN right out of the metal (which explains why they are usually more accurate than the BIOS setting) and who knows where it has placed that "CPU temperature" sensor (the HWInfo authors usually have no idea, they just print the output).
 

·
Premium Member
Joined
·
3,287 Posts
Quote:
Originally Posted by Nuclear Coffee View Post

small bump
You can bump if you like, but the previous post answered the question 100% accurately. Your Nuvoton is simply reporting the wrong temperature. It could be a bad thermal diode, or it could be HWiNFO64 not reading the right register to get the value.

Give HWiNFO64 a few months to get these new Z97 board sensors figured out. You can do your part by posting a message with the same detail on their forum.

Greg

BTW: This isn't the first time the NCT6791D has reported a high CPU temp:

http://www.hwinfo.com/forum/Thread-Solved-ASUS-Z87I-Deluxe-High-Temps

http://www.hwinfo.com/forum/Thread-Asus-H87-pro-High-temps
 

·
Premium Member
Joined
·
5,021 Posts
It's just some sensor somewhere on your board that is wired into that SuperIO chip that's mentioned (that Nuvoton thingy). The name being "CPU" is obviously wrong. That's what HWINFO guessed for it. It was programmed like this, probably from some other board that has the same Nuvoton chip, and the name "CPU" was correct on that other board but isn't for yours. You could try to see if it has a different name in the beta version of HWINFO.

If I would have to guess where a 75 °C temperature on a board's sensor can come from, I'd bet there's a sensor under the VRM heat-sink that supplies VRIN and this is what it is reporting. Another possibility is that 75 number being totally wrong. This sensor reading might need some offset before showing something useful and the HWINFO developer just wasn't told.
 

·
Registered
Joined
·
4,854 Posts
are you running the version 4.40? it is the newest version and it shows it has been updated for Z97 boards and yours is on the list.

edit: I looked at the top of the pic you posted and see that you are running version 4.40
so I would just say it's a issue with the software.
I never trust software to read temps or voltage on my system, I use a volt meter and for temps I have a inferred heat gun.

http://www.guru3d.com/files-details/hwinfo64-download.html

Changes:
•Enhanced sensor monitoring on several ASUS Z97 and H97 models.
•Enhanced sensor monitoring on GIGABYTE Z97 models.
•Fixed reporting of OS UEFI Boot.
•Fixed support of AMD Oland/Mars/Sun/Jet GPUs.
•Fixed reporting of DDR3 memory timings for some modules.
•Added support of IRF IR3580.
•Enhanced sensor monitoring on ASUS MAXIMUS VII series.
•Enhanced sensor monitoring on ASRock Z97 and H97 series.
•Enhanced sensor monitoring on ASRock Z87 and H97 ITX series.
•Added option to hide Welcome screen and Detection progress.
 

·
Premium Member
Joined
·
9,160 Posts

·
HWiNFO Author
Joined
·
1,258 Posts
OK, this needs some explanation
smile.gif

Almost all mainboards have chips on mainboard which read various temperature, voltage and fan values from sensors across the mainboard. They have general inputs for each sensor and in most cases it only depends on the mainboard vendor to which sensor their wire a particular input. So for example Temperature0 (T0) input can be routed to a diode in CPU socket on one mainboard, while on a different model (even from the same vendor) it can be a diode anywhere else. The problem is that there's no universal way (like to read it from the chip) how to know where such an input is connected to. Usually mainboard vendors have tools which based on the particular mobo model know that T0=CPU, etc. Let's say there are 4 temperature inputs (T0-T3) on such a chip, but only two of them are really connected to a diode. These are then reported by such tools. So HWiNFO needs to know about each mainboard model and how those inputs are connected. This is of course a big task, since there are so many different models, and in some cases it's really difficult to determine particular connections. Now what if out of those 4 inputs, all 4 are really used, but only two of them (CPU and MB) reported by tools like ASRock ? I believe this can happen, so that's why HWiNFO reports ALL sensor values when it *thinks* they might return valid data. If it's not sure about a particular value which seems to return correct data, it uses a label which is most common for most mainboards. And this is your case... It seems that the "CPU" values displayed under the Nuvoton sensor is in fact not the CPU temperature. It might be the VR as suggested, or might be invalid as well. But as long as ASRock doesn't report this value and I don't have the schematics of this board, it's almost impossible to know the truth. Thus I suggest to ignore this value and better rely on the internal CPU core values, or CPU (PECI) which should be very close to the internal core values, since this comes straight from the CPU into the Nuvoton chip.

----
I hope this clarifies the issue and helps to understand the problem and how things really work.
If you have any question, just let me know. Either here, or in the HWiNFO thread on this forum: http://www.overclock.net/t/1235672/official-hwinfo-32-64-thread
 
1 - 11 of 11 Posts
Top