On Sun, 2006-11-19 at 11:13 -0500, Tom Horsley wrote: > Here's a weird difference I just noticed between the Xen and > non-xen kernels on FC6: The VCore voltage reported is lower in > non-xen than in xen. > > In gkrellm the sensor in the config file is described as: > > sensor "VCore" "it8712@290/in0" 10000 0 1 0 > > and the sensors program prints: > > it8712-isa-0290 > Adapter: ISA adapter > VCore 1: +1.10 V (min = +1.42 V, max = +1.57 V) ALARM > > The both report 1.10 (by the way - I haven't bothered to > configure the alarm values in the /etc/sensors.conf file, so > pay no attention to the ranges and alarms). > > If I boot the Xen kernel, the value shows up as 1.36 volts > most of the time. The 1.36 value appears to be the correct one > because when I'm in the BIOS hardware monitor, it normally > shows 1.36 volts, and when I boot Windows and run the Biostar > monitor program that came with the motherboard, it also shows > 1.36 volts. > > A weird little oddity. Worth reporting as a bug? (Against > what I wonder?) > > This is a Biostar TForce4 U motherboard running an Athlon 64 > X2 4400+ > HI, Tom, This could be important to you. If I had that reading condition, I would measure the actual voltage to check it. Be aware that voltages out of tolerance on these new processors can produce additional internal heating, resulting in problems either immediate or down the road, because heat is the one enemy of these things (think about the heavy heat sink on your processor). One of the reasons for the lower voltages in the modern processors is to reduce power draw (twinkle twinkle little star, E=I times R. and power equals E*I) so if you change one, the other changes also. Due to the operation of modern digital IC's, current mirrors are used internally in many places. This means that a voltage that is too low can produce a current that is too high, so if E goes to say 90%, i may go to 110%, producing even more heating. Depends on the processor and its internal design. By the way, the instantaneous currents handled by modern processors during a data and address buss change can be as high as 200 Amps. Note that this lasts for pico-seconds, but the heating effects tend to average out, and that is why most modern processors are like toasters. Regards, Les H
