`psi_avgs_work` shows up in PowerTOP

From: Paul Menzel
Date: Thu Jul 23 2020 - 09:58:08 EST


Dear Johannes,


On the Dell Latitude E7250 with Debian Sid/unstable and Linux 5.6.7, running `powertop`, `psi_avgs_work` shows up there with 40 mW to 60 mW.

The battery reports a discharge rate of 7.16 W
The power consumed was 147 J
The estimated remaining time is 1 hours, 31 minutes

Summary: 795.2 wakeups/second, 0.0 GPU ops/seconds, 0.0 VFS ops/sec and 16.8% CPU use

Power est. Usage Events/s Category Description
2.62 W 5.5 ms/s 328.9 Timer tick_sched_timer
817 mW 21.9 ms/s 99.9 Process [PID 519673] firefox
521 mW 3.2 ms/s 65.1 Process [PID 519710] firefox
270 mW 1.9 ms/s 33.8 Timer hrtimer_wakeup
261 mW 0.4 pkts/s Device Network interface: eno1 (e1000e)
162 mW 4.7 ms/s 19.8 Process [PID 520008] /usr/lib/firefox/firefox -contentproc
156 mW 4.0 ms/s 19.1 Process [PID 519728] firefox
146 mW 16.0 ms/s 16.4 Process [PID 72917] /usr/bin/gnome-shell
125 mW 5.8 ms/s 15.0 Process [PID 518973] /usr/lib/thunderbird/thunderbird
121 mW 160.7 us/s 15.2 Process [PID 11] [rcu_sched]
117 mW 5.2 ms/s 14.1 Process [PID 520003] /usr/lib/firefox/firefox -contentproc
100 mW 100.0% Device USB device: Fujitsu Keyboard (Fujitsu)
100 mW 100.0% Device USB device: Jolla (Jolla)
100 mW 0.0% Device Display backlight
100 mW 100.0% Device USB device: USB Optical Mouse (Logitech)
95.4 mW 605.7 us/s 11.9 Process [PID 519018] /usr/lib/thunderbird/thunderbird
91.3 mW 4.0 ms/s 11.0 Process [PID 519906] /usr/lib/firefox/firefox -contentproc
84.9 mW 45.2 ms/s 5.0 kWork intel_atomic_commit_work
80.1 mW 1.1 ms/s 9.9 Interrupt [0] HI_SOFTIRQ
76.6 mW 30.1 us/s 9.6 kWork kfree_rcu_monitor
75.8 mW 58.0 us/s 9.5 kWork kfree_rcu_work
74.5 mW 269.4 us/s 9.3 kWork engine_retire
62.5 mW 1.7 ms/s 7.6 Process [PID 73223] ibus-daemon --panel disable -r --xim
59.6 mW 73.5 us/s 7.5 kWork psi_avgs_work
56.9 mW 8.1 ms/s 6.1 Process [PID 72956] /usr/bin/Xwayland :0 -rootless -norese
54.7 mW 4.7 ms/s 6.3 Interrupt [7] sched(softirq)
54.0 mW 32.1 us/s 6.8 Timer intel_uncore_fw_release_timer
[â]

Is that expected to show? I guess due to the granularity it is?


Kind regards,

Paul