lm-sensors reports incorrect Vcore (off by 1.0V)

[BrianC]

Ubuntu 19.04. At least two applications that tie into and use _sensors_ info (including Psensor and another I cannot find the name of), plus command line "sensors" report a Vcore of ~+0.4 to +0.7V on a Ryzen 7 2700X on an Asrock Taichi X470 mobo. This is WAY below spec, yet the PC is performant.

However, when in the BIOS, Vcore can be seen to be in the ~1.4-1.5V range.

This behavior is consistent with _lm-sensors_ simply reporting a Vcore that is a full 1.0V less than the actual value (e.g. reporting +0.5 instead of +1.5).

[Andreas Hasenack (ahasenack)]

Can you see if this config helps?

https://github.com/rigred/sensor_config/blob/master/X470/ASRock/ASRock-X470-Tachi-Ultimate

lm-sensors upstream doesn't seem to have a config specifically for the x470, just the x370.

[BrianC (brianlowercase)]

 Yes, that code fixed it. Normal Vcore has been resumed on my AsRock X470 Taichi.    Thanks for the quick reply.  [[ You don't have to answer, but any idea why the existing wildcard approach doesn't pick up the Taichi chip "nct6779-isa-0290"   ? ]]
[to have your fix spelled out for others...]In my case, my sensors config file is name/located  sensors3.config in the /etc directory (Ubuntu).To open the file in a text editor, type or paste this into Terminal.sudo gedit   /etc/sensors3.conf    [then key ENTER]
Enter your (super)user password at the prompt.
The edit was done by A) Commenting out the section that used wildcards to name 'nct67..' chips and pasting in your code.    This was done by:(Ctrl-F (find) and enter nct67 as the search value, to get a set of lines that contain nct67).Scroll or click through to identify the line that starts a region of code  (~lines 318-334  of my original sensors3.conf) that
begins: chip "w83627ehf-*" "w83627dhg-*" "w83667hg-*" "nct6775-*" "nct6776-*"  "nct6779-*" "nct6791-*" "nct6795-*" "nct6796-*"
and ends set in8_max  3.3 * 1.10

This region was then turned into a set of comments (you type a # on left handside of each line of this section, in the manner you can see in other parts of the code).  Using commenting ('#') allows you to 'undo' your edit by deleting the hash(#) in case something is messed up.  Deleting the # simply turns the line back into code.  (Or you can save the original unedited sensors3.conf file under a different name before you make any edits).
Then B) The revised section of code from that github link (take care to enter the whole link, reconstructing any linewrap/breaks) was pasted in below this newly-commented region, followed by a file save.

    On Wednesday, June 19, 2019, 02:25:37 PM EDT, Andreas Hasenack <email address hidden> wrote:

 Can you see if this config helps?

https://github.com/rigred/sensor_config/blob/master/X470/ASRock/ASRock-X470
-Tachi-Ultimate

lm-sensors upstream doesn't seem to have a config specifically for the
x470, just the x370.

--
You received this bug notification because you are subscribed to the bug
report.
https://bugs.launchpad.net/bugs/1833127

Title:
  lm-sensors reports incorrect Vcore  (off by 1.0V)

Status in lm-sensors package in Ubuntu:
  New

Bug description:
  Ubuntu 19.04.  At least two applications that tie into and use
  _sensors_ info (including Psensor and another I cannot find the name
  of), plus command line "sensors" report a Vcore of ~+0.4 to +0.7V on a
  Ryzen 7 2700X on an Asrock Taichi X470 mobo.  This is WAY below spec,
  yet the PC is performant.

  However, when in the BIOS, Vcore can be seen to be in the ~1.4-1.5V
  range.

  This behavior is consistent with _lm-sensors_ simply reporting a Vcore
  that is a full 1.0V less than the actual value (e.g. reporting +0.5
  instead of +1.5).

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/lm-sensors/+bug/1833127/+subscriptions

[Paride Legovini (paride)]

Thanks for your report and for the feedback on the configuration file. The best way to have this fixed in Ubuntu is by having the relevant configuration file included in the upstream project [0]. They have several pull requests open for missing hardware, and according to the commit history they do merge them.

I'd try to ask the author of the custom configuration file [1] if he plans to submit his configs for upstream inclusion. Or you may try to open an upstream issue and point the developers to the working config file.

If you do end up filing an upstream bug, please link to it from here. Thanks!

[0] https://github.com/lm-sensors/lm-sensors
[1] https://github.com/rigred/sensor_config/