Hi Sergio, > > Em 05/06/2008, ?s 12:02, Juerg Haefliger escreveu: > >> Hi Sergio, >> >> >>> root at nimbus:~# ~sergio/cpuid >>> eax in eax ebx ecx edx >>> 00000000 00000001 746e6543 736c7561 48727561 >>> 00000001 000006a9 00010800 00000181 a7c9bbff >>> 80000000 80000006 00000000 00000000 00000000 >>> 80000001 00000000 00000000 00000000 00100000 >>> 80000002 20202020 20202020 20202020 20202020 >>> 80000003 56202020 45204149 65687473 72702072 >>> 80000004 7365636f 20726f73 30303031 007a484d >>> 80000005 00000000 08800880 40040140 40040140 >>> 80000006 00000000 00000000 0080a140 00000000 >>> c0000000 c0000002 00000000 00000000 00000000 >>> c0000001 00000000 00000000 00000000 00003fcc >>> c0000002 00003b00 080c0810 08100a13 42000000 >>> >>> Vendor ID: "CentaurHauls"; CPUID level 1 >>> >>> Unknown vendor >> >> Ah very good! Now can you please run cpuid and 'grep "" >> /sys/class/hwmon/hwmon*/device/temp?_input' in close succession a >> couple of times (maybe 5 times with a couple of secs in between) while >> the system is idle and then again while the system is busy (by running >> cpuburn for example)? This should allow me to calculate the correct >> values for the conversion that goes into sensors.conf. > > attached is a 'better' idle, Ive saw the temp2 temperature even lower and > got another batch of sensors readings. > I looked at your files and it doesn't look too promising unfortunately. First, FYI, register eax in the last line of the cpuid dump contains the core temp value in celcius. I.e., c0000002 00003b00 080c0810 08100a13 42000000 -> hex 3b = dec 59 = 59 degrees celcius >From your snapshots, I see that we get a 24C core temp swing (43C to 67C) between idle and after-freeze. You might be able to push the high temp a little higher by running some stress that doesn't freeze your CPU. Maybe a kernel compile. But that's irrelevant for this discussion. Also, temp1 & temp3 never move so I'm assuming they are not connected and can be ignored/disabled. temp2 is the vt1211 internal temp which tracks the CPU temp somewhat. Temp4 is the only other temp input that changes but I don't think it's the CPU temp. Temp1 is the dedicated CPU temp input and temp4 is for thermistors only (I believe). The temp4 values only swing between 2011 and 2022, I don't know if that's reasonable for a thermistor, maybe others on the list can shed some light. But it's certainly not the CPU, maybe an on-board thermistor or nothing at all. My guess is since it's a fanless board, VIA didn't care to connect the CPU thermal diode to the vt1211 more so that the C7 has another thermal diode that can be accessed via SW. No need to take the detour through the vt1211 to get the temp. But of course that doesn't help you with using the CPU temp to control the chassis fan (assuming this is what you want). You have 2 options as I can see: 1) Use temp2 as the input for the vt1211 PWM controller. 2) Use a user-space tool to periodically check the CPU temp and adjust the PWM output accordingly. For option 2) you can either use the cpuid tool to get the temp or you can wait a couple of hours until I post the c7temp driver to the list that gives you temp access via sensors. For manually controlling the PWM outputs, I have a little script that does it which I'll also post shortly. The script basically puts the vt1211 PWM controller into auto mode and drives it to extreme values so that the PWM outputs can be set manually. Hope this helps ...juerg > > > > Sergio Bruder >
