Re: [PATCH v2 0/5] hwmon: k10temp driver improvements

[Jonathan McDowell]

On Sun, Jan 19, 2020 at 07:46:11AM -0800, Guenter Roeck wrote:
> On 1/19/20 2:18 AM, Jonathan McDowell wrote:
> > 
> > In article <20200118172615.26329-1-linux@xxxxxxxxxxxx> (earth.lists.linux-kernel) you wrote:
> > > This patch series implements various improvements for the k10temp driver.
> > ...
> > > The voltage and current information is limited to Ryzen CPUs. Voltage
> > > and current reporting on Threadripper and EPYC CPUs is different, and the
> > > reported information is either incomplete or wrong. Exclude it for the time
> > > being; it can always be added if/when more information becomes available.
> > 
> > > Tested with the following Ryzen CPUs:
> > 
> > Tested-By: Jonathan McDowell <noodles@xxxxxxxx>
> > 
> Thanks!
> 
> > Tested on a Ryzen 7 2700 (patched on top of 5.4.13):
> > 
> > | k10temp-pci-00c3
> > | Adapter: PCI adapter
> > | Vcore:        +0.80 V
> > | Vsoc:         +0.81 V
> > | Tdie:         +37.0°C
> > | Tctl:         +37.0°C
> > | Icore:        +8.31 A
> > | Isoc:         +6.86 A
> > 
> > Like the 1300X case I see a discrepancy compared to what the nct6779
> > driver says Vcore is:
> > 
> > | nct6779-isa-0290
> > | Adapter: ISA adapter
> > | Vcore:                  +0.33 V  (min =  +0.00 V, max =  +1.74 V)
> 
> I see that on all of my boards as well (3900X, different boards and board vendors),
> with temperatures reported by the Super-IO chip sometimes as low as 0.18V (!).
> Yet, there is a clear correlation of that voltage with CPU load.
> I suspect the measurement by the Super-IO chip is a different voltage.
> 
> I don't think there is anything we can do about that without access to more
> information.
...
> The problem with Ken's board is that idle current and voltage are very high.
> The idle voltage claims to be higher than the voltage under load, which
> doesn't really make sense. This is only reflected in the voltage and current
> reported by the CPU, but not by the voltage reported by the Super-IO chip.

I see clear correlation between load/Vcore/Icore/Tdie from your patched
k10temp driver which leads me to believe these numbers are valid for the
2700. Vsoc is fairly consistent and Isoc doesn't vary much either
(6.3-8.1A range over the past 8 hours).

J.

-- 
... "f u cn rd ths, u cn gt a gd jb n cmptr prgrmmng." -- Simon Cozens,
    ox.os.linux