• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Are temp monitor programs wrong?

Soldato
Joined
17 Aug 2009
Posts
18,523
Location
Finchley, London
Someone told me that temperature monitor programs like cpuid, hwmonitor, etc are largely inaccurate, and that there's apparently an article somewhere showing the massive inaccuracies of burntest and CPUID when they thermoprobed the cpu directly. It showed it was around 10-20C lower than the monitoring software was reporting.

Could this be true or is it rubbish? With maximum load on IBT, I get about 45c, so if this was right about monitor programs, that would put my max load at about 25 to 35c. Even with my TRUE and lots of fans, is that a feasible load temperature?
 
I don't think this a hard and set rule as such. I think they CAN be quite inaccurate and often several different programs will report different results. I wouldn't have thought that every processor is 10-20C cooler than reported though since as you mentioned a load temp of 25-35C is pretty unfeasible.
 
Measuring the case of the CPU is completely different to measuring the temp inside the core, there's usually around 15 degC difference, which matches what you've been told. Whoever did that article didn't understand what they were trying to achieve (or should have been trying to achieve).
 
Measuring the case of the CPU is completely different to measuring the temp inside the core, there's usually around 15 degC difference

With HWMonitor, my TMPIN 0 is 26c, (I believe that's the motherboard temp), TMPIN 1 (CPU) is 35c and my core temp is 38c. If by case of the CPU you mean my TMPIN 1, then I never get more than about 3c difference.
 
I tried to google TMPIN and it seems to depend which mobo you have as to what temp it is. To be honest, it's not relevent, the core temp is the one to look at.

With HWMonitor I've just run IBT on my 2500K and have max core temps between 72 and 85 and package temp 88, which doesn't add up at all! package temp 3 degrees more than the hottest core, that's just cr@p.

Real Temp 3.67 and Core Temp 1.0 are both saying 77 to 88 max.

There are problems with some of these temperature monitoring programs, what works for one type of CPU may not be right for another. I go with the hottest.

Measuring the temperature with external equipment is certainly not the way to do it.

Interestingly, both Core Temp and Real Temp show min temps of 24 - 28, HWMonitor says 26 - 29 (package 28!). This would indicate that it's not an offset error but a scaling one.
 
Agree with the above, the best solution is probably to use several temp monitoring utilities and then judge by the worst case scenario. You know you're probably on the safe side of things then.
 
Yes, I've always had weird readings from my core temp but it seems to be something to do with my phenom sensor. In my screenshot, you'll see that my idle temps show core at 18c whiile TMPIN 1 is 32c.
http://img405.imageshack.us/img405/7290/tempsxh.jpg


At load with orthos, say when I had my overclock a while back at 3.8GH, my TMPIN 1 was 42c and core was 26c.



I was told that adding about 20c to whatever the core number makes it about right. So at idle, that would make my core temp 38c not 18c, and at load, 26c becomes 46c. That puts the core temp always slightly above the cpu temp. So, would you say that's the correct way I should read my strange core temps readings, by adding about 20c?
 
there are three things that i can think of that make temp reading inaccurate:

- each program will have its own algorithm to convert the sensor reports to temp readings, so each program could give you a different value - my bios reports almost 10 degrees higher than realtemp for instance!

- the sensors on the chips themselves, we all can say "well, my cpu is 2 degrees cooler than yours" but we have no idea of the inaccuracies of these sensors, if they are +-1 degree, yours could be +1, your friends -1 and so both cpu's are actually running at the same temp

- each manufacturer will use different sensors and so probably needs different algorithms for each manufacturer/sensor type
 
Not sure, but I've heard that the 'core' temps on Phenom II chips have been a bit iffy. I've heard that the actual 'CPU' temp is usually a little higher and is the one you should pay attention to. Maybe this is what you're seeing here?
 
- each program will have its own algorithm to convert the sensor reports to temp readings, so each program could give you a different value - my bios reports almost 10 degrees higher than realtemp for instance!

bios temps aren't the same as idle temps, bios is usually higher because certain power saving features aren't in play.

- the sensors on the chips themselves, we all can say "well, my cpu is 2 degrees cooler than yours" but we have no idea of the inaccuracies of these sensors, if they are +-1 degree, yours could be +1, your friends -1 and so both cpu's are actually running at the same temp

Ambient temperature plays a part too.

- each manufacturer will use different sensors and so probably needs different algorithms for each manufacturer/sensor type

I suspect this is where some of the programs are going wrong, need to check that the software states your CPU and/or mobo is supported.
 
Yes, I've always had weird readings from my core temp but it seems to be something to do with my phenom sensor. In my screenshot, you'll see that my idle temps show core at 18c whiile TMPIN 1 is 32c.
http://img405.imageshack.us/img405/7290/tempsxh.jpg


At load with orthos, say when I had my overclock a while back at 3.8GH, my TMPIN 1 was 42c and core was 26c.



I was told that adding about 20c to whatever the core number makes it about right. So at idle, that would make my core temp 38c not 18c, and at load, 26c becomes 46c. That puts the core temp always slightly above the cpu temp. So, would you say that's the correct way I should read my strange core temps readings, by adding about 20c?

I would suggest you download core temp and real temp, run these alongside hwmonitor. Post the results here so we can take a look :)
 
I would suggest you download core temp and real temp, run these alongside hwmonitor. Post the results here so we can take a look :)

Just took this screenshot. Realtemp only works with my laptop cos it's intel. Doesn't work with AMD.

hwcoretemp.jpg
 
Yeah HWmonitor always says my +12v peaks at 16.26v on my antec trupower 650w but tested and fine. Must be a spike under stress other monitor progs dont pick it up
 
Back
Top Bottom