khali at linux-fr.org
Sat Feb 12 12:46:31 CET 2005
> I think Vcore is measured from pin 'in1', but it produces 'too' good
> stable reading of 1.500 anyway -from day to day. It's not exactly the
> same what I can see in bios, but the reading is quite stable there
> too. The slight difference (1.494 in bios) _could_ be caused by
> slightly different current consumption, maybe?
Yes it could. The LSB of that input voltage channel is worth 0.008V, so
the slightest change could result in the different reading.
> I have tested this both in console, and in a terminal window under X,
> both result always in 1.500. Strange?
Admittedly strange, although not impossible.
> Or am I thinking something in a wrong way again? Below is the output
> of 'sensors' and piece of sensors.conf.
> lm_sensors is version 2.8.7-2, from current FC3 rpm.
> Limits of Vcore are based to VID, which equals to 0. Is it a variable
> that needs to be se manually in the file, or should it be
> automatically readable from somewhere.
VID is read from the chip and should not be 0. There are two reasons it
1* The motherboard manufacturer did not wire the VID input pins.
Unlikely with Intel.
2* You use a recent kernel and a not-so-recent lm_sensors. For a
2.6.9-rc1 or higher kernel, you need lm_sensors 2.8.8 or higher.
> Adapter: SMBus I801 adapter at c400
> ERROR: Can't get alarm mask data!
> V1.5: +1.471 V (min = +1.42 V, max = +1.58 V)
> VCore: +1.500 V (min = +0.00 V, max = +0.00 V) ALARM
> V3.3: +3.403 V (min = +3.13 V, max = +3.47 V)
> V5: +5.156 V (min = +4.74 V, max = +5.26 V)
> V12: +12.313 V (min = +11.38 V, max = +12.62 V)
> CPU_Fan: 3170 RPM (min = 4000 RPM) ALARM
> 2881 RPM (min = 0 RPM)
> 2884 RPM (min = 0 RPM)
> CPU: +27.00°C (low = +10°C, high = +50°C)
> Board: +25.00°C (low = +10°C, high = +35°C)
> Remote: +27.00°C (low = +10°C, high = +35°C)
> ERROR: Can't get PWM1 data!
> ERROR: Can't get PWM2 data!
> ERROR: Can't get PWM3 data!
Errors on PWM would also be explained by a "too" recent kernel.
I'd invite you to try a newer version of lm_sensors, that would
certainly help. If this isn't an option, then you can temporarily do the
> chip "lm85c-*" "adm1027-*" "adt7463-*" "lm85-*" "lm85b-*"
> # PWM Outputs
> ignore pwm1
> ignore pwm2
> ignore pwm3
> set in1_min 1.500 * 0.95
> set in1_max 1.500 * 1.05
That way the errors on PWM will ge hidden and the limits for Vcore will
be set correctly (after you run sensors -s).
More information about the lm-sensors