CPU temperature differences with several monitor programs

TJ
TJ
Joined: 11 Feb 05
Posts: 178
Credit: 21,041,858
RAC: 0
Topic 197033

I have and 6 old quad core which could fit 1 or 2 GPU's for crunching. But first I was checking its power consumption. That is not bad. I let it crunch via CPU and saw strange temperature readings.
TThrottle: 40-47°C idle and ~63° crunching
Core temp: 27-33°C idle and 42-47° crunching
Real temp: 30-37°C idle and 60-65° crunching

That makes it hard to say which program is right. Its a Dell MOBO and the program about info about it is not running. I tried updating it but that does not work. Its an old LGA775 socket.

The question is: does anyone have experience with which program has the most reliable temperature readings? Thanks.

Greetings from
TJ

Gavin
Gavin
Joined: 21 Sep 10
Posts: 191
Credit: 34,802,233,901
RAC: 931,124

CPU temperature differences with several monitor programs

It all depends. Most of the monitoring software's are not directly comparable IMO...

It would be helpful to know what quad you are running, how you are running it e.g all cores @ 100%, 100% of the time etc. Room temperature and heatsink/fan combo, are you overclocking? Is the temp monitoring software taking its reading from the chip or the motherboard sensor? Which then begs the question, how accurate is that sensor?

Coretemp has always been a favourite of mine and has some nice features including overheat protection (which works ;-)) and (for me) returns believable figures.

I can't really comment further on your particular quad core as you have chosen to hide your computers, but to give you an indication from my 3 Intel Q6600 quad core hosts (all overclocked to around the 3Ghz mark). I was getting load temps (using Coretemp) of mid to upper 60's Celcius with all cores loaded at 100%, 100% of the time with a room temp. hovering around 26,27 Celsius. Idle temp approx. mid 40's with aftermarket HSF's and open cases (side panels removed).

archae86
archae86
Joined: 6 Dec 05
Posts: 3,010
Credit: 4,859,198,832
RAC: 3,344,310

A pretty basic problem is

A pretty basic problem is that Intel's per-core temperature sensors were developed to serve a reliability purpose, and were never intended to provide an end-user temperature. There is actually quite a lot of random scatter in readings from die to die and from core to core within the same die due to what amount to fixed offsets arising from minor manufacturing variation.

Another problem is that many motherboards provide an external CPU temperature sensor--not to be confused with the within-die sensors. Regarding precision, absence of missing counts, and random sample-to-sample variation these are often quite a lot better than the on-die sensors--but that does not get around the problem that they are not measuring the same thing, so won't even have the same time-dependent waveform. It may be that in making your comparisons you are in some cases confounding a motherboard CPU temperature sensor report with on-die reports.

A relatively simple problem is that the temperature reported in a digital format by the on-die sensors is not on any single standard scale, but is the difference from TJMAX, where TJMAX is a reference temperature specific to a given product. As this whole scheme was not intended to provide a customer thermometer, the TJMAX for a given product is not always known, and one common source of reporting differences among programs is that for certain products they guess different TJMAX values. An additional source of noise is the widespread misconception that TJMAX for a die is in fact in any sense a specified maximum temperature for that die.

The programs make some efforts to let you compensate for some of these problems. The betters ones often display the value of TJMAX they are currently assuming somewhere, and some let you adjust it directly. Also, some programs let you introduce calibration offsets separately for each core-reporting sensor. (as a footnote, be advised that only physical cores have independent sensors, so if someone runs a quad-core hyperthreaded and sees eight core temperatures reported, there are really only four independent ones).

There is more to it than this, but that should give you a start. As a personal choice, I'm used to using Speedfan, and I do introduce core offsets to null out the spurious core-to-core differences. There may well be choices available which would better suit your purpose.

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,211
Credit: 136,093,955
RAC: 50,005

RE: A pretty basic problem

Quote:
A pretty basic problem is ...... choices available which would better suit your purpose.


Wow! What a great explanation. Thanks archae86 :-)

I have a quad core hyperthreading to eight, and I always thought that a particular one of the four was 'bad' simply because it always ran hotter when idle and rose more in temperature than the others when active. Now I see that it could be only a measurement issue - the chosen basepoint may vary and the slope too. For that matter, I guess, linearity may also be ( wrongly ) assumed.

Cheers, Mike

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

dunx
dunx
Joined: 13 Aug 10
Posts: 119
Credit: 53,470,527
RAC: 0

But it is really poor Asus's

But it is really poor Asus's own monitor for the Maximus 6 Impact is 20 degrees off target !

Worse still it is 20 degrees too low !

dunx

Mumak
Joined: 26 Feb 13
Posts: 325
Credit: 1,932,011,719
RAC: 0

Regarding internal (DTS) CPU

Regarding internal (DTS) CPU temperatures, you might read this:
http://www.hwinfo.com/forum/Thread-CPU-Core-temperature-measuring-via-DTS-Facts-Fictions
I wrote that a few years ago, but it's still valid. Many people don't realize how important the accuracy is, and that in many cases the values they get are just bogus.
No, I'm not advertising my own tool/site here, it's full on topic ;-)

-----

Milan
Milan
Joined: 8 Sep 20
Posts: 5
Credit: 4,336,982
RAC: 16,953

What is the best temperature

What is the best temperature monitoring program and what those terms means? I am using HWinfo64

What should i look for when i monitoring voltages and temperatures?

Where i can see VRM temperature, what is VID?

Is it Cores VIDs voltages thing what i should look for when i check voltages on my CPU?

Etc.

Sorry if i ask too much questions but i don't know too much about computers so i try too learn something.

I don't overclock or downclock my computers but i use ''Power saver mode'' to lower temperatures and voltages and i don't know is it good thing to use ''Power saver mode'' in terms of longevity for pc hardware, that's why i want to monitor my pc and to know what to monitor :)

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 370
Credit: 587,192,675
RAC: 2,912,165

Milan wrote: What is the

Milan wrote:

What is the best temperature monitoring program and what those terms means? I am using HWinfo64

What should i look for when i monitoring voltages and temperatures?

Where i can see VRM temperature, what is VID?

Is it Cores VIDs voltages thing what i should look for when i check voltages on my CPU?

Etc.

Sorry if i ask too much questions but i don't know too much about computers so i try too learn something.

I don't overclock or downclock my computers but i use ''Power saver mode'' to lower temperatures and voltages and i don't know is it good thing to use ''Power saver mode'' in terms of longevity for pc hardware, that's why i want to monitor my pc and to know what to monitor :)

Hello Milan, welcome to Einstein!!

First off, don't be sorry for asking too many questions.  That is what we're here for.  And it is good that you want to watch what is going on with your computers, and if you have a computer that you bought new (not used) personally I don't think you need to to use it in "Power saver mode" unless you need to reduce power consumption from the wall outlet to reduce your electricity usage.  I have an old (11+ years) computer that I have always run in "Performance" mode and it is still working as it always had.

I assume that you are running Windows because you mentioned that you are using HWiNFO64 to monitor your computer.  That is one of the best and most accurate programs for monitoring hardware, if not the best.  You can look at anything you want (so to speak) with HWiNFO64.  Here is a picture from one of my computers running Windows10 and HWiNFO64:

As you can see, you can get VRM temps in it, as well as the Chipset temperature and fan speed, GPU info, etc.

I do not know what computer you are using, or what motherboard, but if you look at your BIOS/UEFI you may find settings for many things, including VID, and it will tell you what it means.

Have fun, and don't be afraid to ask more questions.  That why it's here.

George

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 1,319
Credit: 2,372,911,463
RAC: 8,189,788

VID stands for Voltage

VID stands for Voltage Identification.  It is basically a lookup table programmed into cpus and gpus that index voltages required to operate at a specific clock frequencies.

As you slide up in clock frequencies that a device operates at, the device asks for the correlated voltages from the VRM's needed to run that clock frequency.

And no VID table is identical in one device to another.  It is all programmed into the silicon based on silicon or device quality.

And VID does not equal Vcore. You should look at Vcore to view what your device is actually running at at any one time.

 

Milan
Milan
Joined: 8 Sep 20
Posts: 5
Credit: 4,336,982
RAC: 16,953

This is in ''Power saver

 This is in ''Power saver mod''

This is ''AMD balanced''

AMD balanced

 

CPU Ryzen 5 1600AF and ASUS a320M-K motherboard. That was a prebuilt configuration and it was new also cheap. Temperatures in ''Power saver mod'' are low and in ''AMD balanced'' can go higher then this, in mid 60s and under CPUID stres tes around 70C. PC case is also cheap without fans and i can't upgrade it until warranty expire.

I have two more PC's  and one is like brand new but old :) 9w CPU so it's running cold and it could run on full load all day long. Other PC was running hot and now it turn of sometimes after full load for extend period of time. I don't know when but it was set to run at his full potential in amd catalyst and that's why i am careful now :)

VRM's are probably ''Temperature 6'' this is a cheap mobo...

Vcore are CPU Core  Voltage ( Voltage provided via voltage regulator) ?

Sry because image upload but it is on ''flickr'' i didn't upload image on forums for a long time :) i will try to post image properly later this day.

Thanks for answering.

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 1,319
Credit: 2,372,911,463
RAC: 8,189,788

Yes, Temp 6 has traditionally

Yes, Temp 6 has traditionally been the VRM regulator temps on ASUS boards.

 

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.