Hardware sensors in Linux
So I setup lmsensors in Linux the other day and found some information I didn't even have with my MSI core Cell utility in Windows, namely the individual core temperatures of my AMD X2 processor. here is the output of sensors after running BOINC/SETI for a while:
john@beast:/usr/local/BOINC$ sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp:
+54°C
Core1 Temp:
+54°C
w83627thf-isa-0290
Adapter: ISA adapter
VCore: +1.56 V (min = +0.70 V, max = +1.87 V)
+12V: +12.65 V (min = +7.90 V, max = +0.24 V) ALARM
+3.3V: +3.23 V (min = +1.15 V, max = +0.13 V) ALARM
+5V: +5.09 V (min = +0.03 V, max = +2.03 V) ALARM
-12V: -12.28 V (min = +2.77 V, max = -14.75 V) ALARM
V5SB: +5.16 V (min = +0.43 V, max = +1.72 V) ALARM
VBat: +3.12 V (min = +3.58 V, max = +1.09 V) ALARM
fan1: 0 RPM (min = -1 RPM, div = 2) ALARM
CPU Fan: 5818 RPM (min = -1 RPM, div = 8) ALARM
fan3: 4017 RPM (min = 10546 RPM, div = 2) ALARM
M/B Temp: +43°C (high = +34°C, hyst = +0°C) sensor = thermistor ALARM
CPU Temp: +45.5°C (high = +80°C, hyst = +75°C) sensor = thermistor
temp3: -98.5°C (high = +80°C, hyst = +75°C) sensor = diode
vid: +0.800 V (VRM Version 2.4)
alarms: Chassis intrusion detection ALARM
beep_enable:
Sound alarm enabled
and after idling for a while:
john@beast:/usr/local/BOINC$ sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp:
+37°C
Core1 Temp:
+36°C
w83627thf-isa-0290
Adapter: ISA adapter
VCore: +1.55 V (min = +0.70 V, max = +1.87 V)
+12V: +12.77 V (min = +7.90 V, max = +0.24 V) ALARM
+3.3V: +3.25 V (min = +1.15 V, max = +0.13 V) ALARM
+5V: +5.12 V (min = +0.03 V, max = +2.03 V) ALARM
-12V: -12.28 V (min = +2.77 V, max = -14.75 V) ALARM
V5SB: +5.16 V (min = +0.43 V, max = +1.72 V) ALARM
VBat: +3.10 V (min = +3.58 V, max = +1.09 V) ALARM
fan1: 0 RPM (min = -1 RPM, div = 2) ALARM
CPU Fan: 6026 RPM (min = -1 RPM, div = 8) ALARM
fan3: 4115 RPM (min = 10546 RPM, div = 2) ALARM
M/B Temp: +41°C (high = +34°C, hyst = +0°C) sensor = thermistor ALARM
CPU Temp: +37.0°C (high = +80°C, hyst = +75°C) sensor = thermistor
temp3: -98.5°C (high = +80°C, hyst = +75°C) sensor = diode
vid: +0.800 V (VRM Version 2.4)
alarms: Chassis intrusion detection ALARM
beep_enable:
Sound alarm enabled
basically the same numbers I get in windows except the individual cores I don't get at all!
but in windows at idle the MSI utility says about 34 degrees on CPU at idle and anywhere from 43-46 at load. I don't know what temp3 is at -98 but it must not be active on my mobo lol. Also the mobo temp in the MSI core utility on windows doesn't work.. though I haven't updated that software since I installed windows when I got this thing. Interesting I thought ;-)
![[]](/modules/ecommerce/cart/images/cart_empty.png "View your cart items"){kind=spacer}
That's interesting how the Core Temp is about the same as each individual core while idling, but when put under load, the individual cores are about 10C hotter than the reported Core Temp. I'm willing to bet that there is an actual thermistor near the CPU socket that reports the Core Temp, and that the actual die diodes are being used for the two Core readings.
Actually, I'm not all that surprised. The heat is being pulled out of the surrounding area quick enough that the thermistor isn't heating up to the point of the cores. There is loss at several levels, but I didn't think it would be quite so high. I would guess that, according to these results, removal of the integrated heat-spreader on the CPU could decrease overall temps by about 5C. I wouldn't try it, but it looks like there's a good bit of loss there.
Thanks JB!
Interesting thoughts. Also I think your conclusions are accurate based on some other observations. When I start up SETI or something equally intensive the individual core readings jump up to 45 from the idle 33 IMMEDIATELY and then gradually rise to 50+ like you can see after it's been running for a while, and when I turn off SETI they drop down about 10 degrees or so immediately. The CPU temp reading adjusts a lot more slowly. So I think it's pretty definite it has the diode on each core and then a separate thermistor for the CPU socket or something similar. ;-)
About the loss from the heat spreader.. ya I think I'll leave that on even if I could get it to drop 5 degrees, not like I can OC it anymore anyway w/o a different mobo that let me increase the vcore some more.