cpu temp monitoring

process

New Member
Hi, I had a quick browse over the forum looking for an answer to my question, but seems this is build dependant?

Clocking a fx8350 has given some high temps and I'm not really sure which temps to read and be more concerned about. CPU has read up to 69c while CPU0 is a lot less. Also unsure of what T2 is under rog.

I know the threshold of the 8350 is around 62c, but which temp does this refer to cpu or cpu0?







Thanks for any help.
 

Martin

HWiNFO Author
Staff member
The "CPU 0" temperature under AMD processor section comes straight from the CPU package sensor. Unfortunately later AMD CPUs seem to have this sensor extremely inaccurate, especially at lower temperatures. They should provide more precise values when the temperature gets closer to the critical point.
The "CPU" temperature under IT8712F comes most probably from a sensor (diode) placed near the CPU socket, however only the board designer (ASUS) knows this exactly.
So it's hard to make a precise judgment, my personal opinion is that the "CPU" value should be more precise (I don't trust the internal AMD CPU sensors in later CPUs at all). You might want to verify this with an ASUS monitoring tool.
 

process

New Member
Martin said:
The "CPU 0" temperature under AMD processor section comes straight from the CPU package sensor. Unfortunately later AMD CPUs seem to have this sensor extremely inaccurate, especially at lower temperatures. They should provide more precise values when the temperature gets closer to the critical point.
The "CPU" temperature under IT8712F comes most probably from a sensor (diode) placed near the CPU socket, however only the board designer (ASUS) knows this exactly.
So it's hard to make a precise judgment, my personal opinion is that the "CPU" value should be more precise (I don't trust the internal AMD CPU sensors in later CPUs at all). You might want to verify this with an ASUS monitoring tool.
Thanks for your reply. 8350 owning clockers agree with it being core temp and say:

Cpu 0 is your core/package temp
ignore cpu temp really, unless its skyhigh
its generally 10C above what your cpu 0 will be
 

Martin

HWiNFO Author
Staff member
I'd say ignore the CPU temp as well, but I know how unreliable the internal core/package temperature in later AMD CPUs is.. And AMD has acknowledged this issue, but there's no fix for it...
 

trodas

Well-Known Member
That is a fix for it... Intel CPU :) But that is not the answer what one could hoped for...

Also the ATI PM2 voltages seems a bit off :) 897 - 916 volts? What he is powering his rig with? Portable nuclear plant? :D
(beware of Fukushima!)
 

Martin

HWiNFO Author
Staff member
ATI PM2 is a tough sensor, in many cases not used or calibrated. I think this is such case too, so best is to ignore it.
 

trodas

Well-Known Member
You are right - this is obviously far from reality :) 10kV jump thru like 10mm of dry air, so... impossible to create PCB for that :)

But hey, todays I seen miracle! Temperature of VRM components on my PNY 6800GT hit -9999°C! :)
I claim a WR in many cases:
1) coldest running VRM
2) reaching lowest temperature ever
3) making all the scientist look dead stupid by their fake claims of absolute zero temp a bit under -273°C *
4) fastest cooling down from +68°C to -9999°C in just two seconds! :)

Heheheheh!

/me *) of course know, that this is just a glitch in HWiNFO, but funny one! Wanted to take a screenshot, but it never ever happen again! Damn... no WR today :)

 

angrybear168

New Member
Martin said:
I'd say ignore the CPU temp as well, but I know how unreliable the internal core/package temperature in later AMD CPUs is.. And AMD has acknowledged this issue, but there's no fix for it...
Hi I was wondering if this problem has been fixed? if not fixed which temperature should I look at for maximum? Thank you! I have AMD 8320。
 

Martin

HWiNFO Author
Staff member
The problem of AMD internal CPU temperatures cannot be fixed in software, it's a hardware flaw. Some tools seem to add an offset to the the values read, but I believe this is not correct.
You might rely on an external CPU sensor placed close to the socket, but this is mainboard-dependent. Though that won't provide the internal CPU temperatures.
 
Top