linux-kernel - Fwd: power savings in kernel 5.5 over 5.4

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Fri, 13 Mar 2020 21:38:55 +1000
From:   brett hassall <brett.hassall@...il.com>
To:     linux-kernel@...r.kernel.org
Subject: Fwd: power savings in kernel 5.5 over 5.4

Hi

I'm seeing power savings of up to 50% in kernel 5.5 compared to 5.4.

I wrote a small program to capture the current_now value from
/sys/class/power_supply/BAT0 every 5 sec. The program buffers these
values for 10 min (until the screen blank kicks in) then writes them
to a file. Only current_now is captured to minimize measurement impact
(voltage does vary but only by around 10% compared to 300% for
current).

Attached are the plotted values from Ubuntu kernel 5.4.0-14 (based on
5.4.18) and  Ubuntu mainline kernel 5.5 rc1. The method used is: boot
the system, startx, start terminal, wait for the system to settle, run
10 min capture.The orange curve is the 5.4 kernel and the blue is 5.5.
I have run many such captures and the 2 curves provided are
consistently reproducible. The idle current with 5.5 is clearly more
constant.

The partition used for the generating the captures is Ubuntu 20.04,
minimal install, Openbox/Tint2 "desktop" booted with
nouveau.modeset=0. Laptop is Dell P5530. Power save is enabled on
sata/pci/usb devices.

Can anyone suggest what change to 5.5 would be responsible for the
more constant idle power usage and whether it would be feasible to
backport this to the long term kernel 5.4 ?

Thank you
Brett

Download attachment "current_plots.png" of type "image/png" (84207 bytes)