Interpreting FX 8350 and Motherboard Temperatures

[Perfect_Hand]

First time poster, here.

I'm preparing a brief report for my IT department on the differences in power consumption between several different hardware configurations. I'm using a Kill A Watt meter while running Prime95's small FFT stress test to measure power consumption under load.

I figured I'd track temperatures as well since I was running stress tests on various systems. It's not pertinent to the final report, but it wouldn't hurt to have the information. My problem is that I am having trouble interpreting the readouts for the FX 8350 configuration. If it helps, the board is a Gigabyte GA-78LMT-USB3.

CPUID Hardware Monitor was not providing data that I could easily interpret so I decided to run Speedfan and Hardware Info as well. I've done some Google Fu to help understand these figures, but I can only find mixed and often conflicting information. In reference to the attached screenshot, my conclusion is that the CPU IHS temperature is 52 degrees while the socket temperature is 58 degrees. To me, this relationship seems reversed from what I've seen with Intel systems. Additionally, Speedfan and Hardware Monitor report TMPIN02 as 67 degrees. If this is the North Bridge, it seems odd that it would be warmer than the CPU and socket.

This is an office environment so I'm running everything stock.

Can anyone shed some light on these figures for me? I'd greatly appreciate it.

 

[SlowSpyder]

I don't think there is a way to know for absolute certainty what your AMD's FX temp is via software tools. Is it likely that your system will be fully loaded, all eight cores? I'd just simulate a typical load and go with the reported numbers from the Kill-o-Watt and reported temps. They should be in the ballpark, though probably not 100% on.

 

[ShintaiDK]

You cant use AMDs temperatures to anything but the distance to the thermal margin.

 

[janeuner]

IIRC, AMD's 32nm temp sensors are only accurate to +/- 20C, and the error seems to be always cooler than reality. That's why the oft-quoted max safe temp is only 70C, even though the process breakdown temp is ~95C.

Side note, the reported package power is incredibly sad. The difference between what the engineering team thought they would get and what the actually got is off by a factor of two.

 

[AtenRa]

IIRC, AMD's 32nm temp sensors are only accurate to +/- 20C,

Clearly you mistaken.

 

[janeuner]

I wish I was. My own chip is currently idling at a supposed 8C, which is pretty amazing considering the ambient temp in the room is 20C.

 

[thilanliyan]

I had an FX6100 before and it never read below ambient. I was using Coretemp.

 

[SlowSpyder]

I wish I was. My own chip is currently idling at a supposed 8C, which is pretty amazing considering the ambient temp in the room is 20C.

I have Coretemp installed, and mine does this. Below a certain temp the reading is all over the place and completely worthless. Once I give the CPU some load it starts reading more in line with what you'd expect. For whatever it is worth, my machine has shut down more than a few times as I ran up to 70C (according to Coretemp) while I was beating on my CPU. :sneaky:

 

[janeuner]

From the CoreTemp FAQ:

5. Why is the temperature of my FX, Phenom, Athlon based processor lower than the ambient temperature?
Starting with the Phenoms, AMD's digital sensor no longer reports an absolute temperature value anymore, but a reading with a certain offset, which is unknown. It is estimated that this offset is between 10 - 20c.