[lm-sensors] initial fintek f71882fg driver, test please
Hans de Goede
j.w.r.degoede at hhs.nl
Sat Jul 7 15:26:09 CEST 2007
Walt H wrote:
> Goede, J.W.R. de wrote:
>> On Wed, 20 Jun 2007 18:28:49 -0500
>> Walt H <walt_h at lorettotel.net> wrote:
>>> This was tested on an MSI P965 Platinum MB. I installed
>>> the svn checkout of libsensors3 and copied the section
>>> from the f71805 to modify for this chip. The set
>>> statements for the chip fail, so I'm not able to set
>>> hysteresis or critical values on temperatures yet. Also,
>>> on my MB, the BIOS hardware monitor only shows the CPU
>>> Vcore, +5 VSB, +5, +12, NB and DRAM voltage, so I can't
>>> vouch for the other inputs.
>> Well, the 3.3V, 3.3VSB and Vbat inputs are all internally
>> connected in the chip, so they should be fine. That still
>> leaves us one voltage short though.
>> The comments from the f71805
>>> indicate that in0 (+3.3) should be internally scaled and
>>> need no adjustment, however, if that input is correct it
>>> appears to be scaled in half by default, so I multiplied
>>> it by 2.
>> Ahha, if you would have looken at the epox sensors.conf I
>> attched and am using for tested you would have seen that
>> I'm scaling in0 there by a factor of 2, and a comment not
>> to change this as its internally scaled. I guess the f71805
>> driver is doing the scaling in the driver already. Which
>> makes sense. I'll modify my driver todo the same for in0
>> and in6 and in7.
> Sorry, missed that.... Figured F71805 would be close.
>> Also Vcore is most likely connected to in1, as in1 is the
>> only one which can cause voltage alarms.
> Makes sense.
>> I think using the config below should yield better results,
>> can you try this please? ALso please write down your BIOS
>> readings till the last diget, assuming you have a decent
>> powersupply the readings should be pretty stable, so maybe
>> we can compare the last digits to find out exactly whats
>> what. Because currently in2, in3 and in4 are all reading
>> about 1V unscaled, which when using normal scaling factors
>> could be any of: 1V 2V or 5.2V when scaled, so the lables
>> currently apointed to them below are somewhat random ATM.
>> chip "f71882fg-*"
>> # Temperature
>> label temp1 "CPU"
>> label temp2 "System"
>> ignore temp3
>> ignore temp4
>> # Fans
>> label fan1 "CPU"
>> label fan2 "System"
>> ignore fan3
>> ignore fan4
>> # Voltage
>> label in0 "3.3V"
>> label in1 "Vcore"
>> label in2 "Vram"
>> label in3 "5VSB"
>> label in4 "+5V"
>> label in5 "12V"
>> label in6 "??"
>> label in7 "3VSB"
>> label in8 "Batery"
>> # never change the in0, in7 and in8 compute, these are
>> hardwired in the chip!
>> compute in0 (@ * 2), (@ / 2)
>> compute in2 (@ * 2), (@ / 2)
>> compute in3 @*(1+200/47), @/(1+200/47)
>> compute in4 @*(1+200/47), @/(1+200/47)
>> compute in5 @*(1+200/20), @/(1+200/20)
>> compute in7 (@ * 2), (@ / 2)
>> compute in8 (@ * 2), (@ / 2)
> This looks good Hans. The Vcore is seems to be spot on now. It does
> range, as expected due to freq. scaling, but it matches up with what
> I've seen in the BIOS. All the rest look good as well. The ?? (in6)
> sensor appears to be NB as that is what is set in the bios under the
> "overclocking" menu.
> Also, tried booting without pnpacpi turned off and it worked just fine.
> Current output is as follows:
> Adapter: ISA adapter
> 3.3V: +3.31 V
> Vcore: +1.14 V
> Vram: +1.98 V
> 5VSB: +5.05 V
> +5V: +5.00 V
> 12V: +12.23 V
> ??: +1.26 V
> 3VSB: +3.31 V
> Batery: +3.17 V *** (spelling: S/B Battery)
> CPU Fan: 1066 RPM
> System Fan: 1469 RPM
> CPU: +39.0 C (high = +85.0 C, hyst = +4.0 C) sensor =
> temp1_crit: +100.0 C (hyst = +4.0 C)
> System: +34.0 C (high = +85.0 C, hyst = +4.0 C) sensor =
> temp2_crit: +100.0 C (hyst = +4.0 C)
Thanks for testing,
I've added this config for the MSI P965 Platinum to the wiki, see:
More information about the lm-sensors