Hi Mathias, On Sat, 22 Feb 2014 02:29:48 +0100, Mathias Gerber wrote: > On 21.02.2014 11:21, Jean Delvare wrote: > > What makes you think this is Vtt and not Vcore? Almost all boards > > have Vcore on in0. > The actual reading is 1.07V, and if I change the CPU Vtt Voltage in > the BIOS, the value also changes according. > The same feature was in discussion on the AIDA forums [1] for almost > the same mainboard. OK. > >> compute in1 1.649*@,@/1.649 # multipliers provided by Martin > >> Malik author of hwinfo > > Couldn't you figure it out from the BIOS? > As I installed the mainboard I found the factor in the wiki [2] from a > board with the same hwmon chip. With this my BIOS and lm-sensors > readings matched (BIOS: 3.304, lm-sensors -u: 3.305) So I did no > further investigations. > > But now since you asked my I tried to accept the challenge ;-) Sadly > I'm very confused now.. I tried to find as many 3.3V BIOS readings > from the same or similar boards and found the following: Beware of "similar boards". We've seen boards with almost the same name that had slightly different monitoring setups. Sometimes even different hardware revisions of the same board can differ in the details. I seem to recall I've even seen once a BIOS update which changed one scaling factor as apparently is was wrong originally. > U [V] U_Delta [V] Ref URL > 3.284 > 3.304 0.02 > 3.324 0.02 > 3.344 0.02 [3] > 3.363 0.019 [4] > 3.383 0.02 > 3.423 0.04 > > Very confusing measurements, especially the step after 3.363V upwards! > With all these measurements I assumed a 0.02V step in the reading and > calculated the factor (20 / 12) which leads to an too high reading > (3.340 instead of 3.304V) > With an assumed step of 0.019V the reading is to low (3.173V instead > of 3.304V) > Calculated reverse the BIOS reading and the unscaled lm-sensors output > (3.304*12/2.004) I get an (19.78 / 12) factor which leads me to an > almost corresponding reading to the BIOS value in sensors (3.303V) You should keep a lot of decimals when searching for the scaling factor. 3.304*12/2.004 is really 19.784431, not 19.78. And 19.78443 / 12 is closer to 1.649 (Martin Malik's value) than 1.648 (which leads to 3.303 V, not matching the BIOS.) > But the scale seems to have an strange stepping anyway.. Have you seen > this before? Yes, we've seen this before, in particular on Gigabyte boards. See the +12V scaling factor here for example: http://www.lm-sensors.org/wiki/Configurations/Gigabyte/GA-945GCM-S2L Martin Malik's 1.649 actually looks good to me, as it maps register values to exactly all the BIOS values you listed above. So let's stick to it. > >> set in7_min 3.00 * 0.95 set in7_max 3.00 * 1.05 > > 3VSB is normally 3.3V so these limits look wrong. > You're right, my fault. > > set in7_min 3.3 * 0.95 > set in7_max 3.3 * 1.05 > > Actual reading here in7_input: 3.384 OK, fixed in the wiki. > > Does the it87 driver say anything about voltage mapping when you > > load it? If in7 us 3VSB it should already be labeled properly. > Yes it is. OK, fixed. Are there other pre-labelled voltage inputs? > >> label temp1 "PCH temp??" # shows always 33°C set temp1_min 10 set > >> temp1_max 60 > I monitored these temperatures after a cold boot and this value seems > to match the "System temperature" in the BIOS health status. > > label temp1 "System temp" > set temp1_min 10 > set temp1_max 60 OK, wiki updated. > >> ignore temp2 label temp2 "MB temp??" # shows always 25°C set > >> temp2_min 0 set temp2_max 60 > Checked this temp again. Found that on the GA-Z77X-D3H this is PCH > temperature. On this board it has also an BIOS reading for it, I dont. > > > Current version of the configuration file is here: Thanks for your > > contribution. > Thank you! -- Jean Delvare Suse L3 Support _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
