If you have time you could implement an experiment in piefed or lemmy to test how it feels.
Would this actually prevent the decay, or just slow it down?
Is there a simpler/better mechanism I’m missing?
I guess there are sociology books written about it, but I don't know any.
I have ruled out the RAM issue by running as simplified test as possible: spotify + stress-ng and easy effects not running
free -h
total used free shared buff/cache available
Mem: 13Gi 7,8Gi 851Mi 166Mi 5,9Gi 5,7Gi
Swap: 35Gi 2,3Gi 33Gi
iostat
Linux 7.1.3-ogc3.4.fc44.x86_64 (bazzite) 17.07.2026 _x86_64_ (12 CPU)
avg-cpu: %user %nice %system %iowait %steal %idle
10,04 0,01 2,64 0,11 0,00 87,20
Device tps kB_read/s kB_wrtn/s kB_dscd/s kB_read kB_wrtn kB_dscd
nvme0n1 0,03 0,67 0,00 0,00 11939 0 0
nvme1n1 108,97 1720,07 2211,60 2309,40 30620304 39370413 41111372
zram0 145,99 105,04 478,97 0,00 1869980 8526568 0
I have started to dig more into the perf, trace-cmd and cyclictest and finally managed to run all of the in the same time to capture the same high delay event. If the arch forum is correct then pipewire uses data-loop.0 process for realtime processing. It got SCHED_RR, so it seems correct as according to man pages it would preempt the SCHED_OTHER that stress-ng runs in.
Interesting thing I (and a council of clankers) noticed in trace_cpu_4.txt is that the interrupt didn't fire at specified time, and when it did, scheduler immediately preempted stress-ng to run the cyclictest as it's SHED_FIFO.
My current theory is some sort interrupt problem
trace-cmd report > trace.txt sed -n '2682652,2684299p' trace.txt > trace_extract.txt cat trace_extract.txt | grep -i "\[004\]" > trace_cpu_4.txt
logs:
ps -eo pid,class,rtprio,ni,pri,comm -L | grep -i 'rr\|ff\|dln'
19 FF 99 - 139 migration/0
21 FF 50 - 90 idle_inject/0
24 FF 50 - 90 idle_inject/2
25 FF 99 - 139 migration/2
30 FF 50 - 90 idle_inject/4
31 FF 99 - 139 migration/4
36 FF 50 - 90 idle_inject/6
37 FF 99 - 139 migration/6
42 FF 50 - 90 idle_inject/8
43 FF 99 - 139 migration/8
48 FF 50 - 90 idle_inject/10
49 FF 99 - 139 migration/10
54 FF 50 - 90 idle_inject/1
55 FF 99 - 139 migration/1
60 FF 50 - 90 idle_inject/3
61 FF 99 - 139 migration/3
66 FF 50 - 90 idle_inject/5
67 FF 99 - 139 migration/5
72 FF 50 - 90 idle_inject/7
73 FF 99 - 139 migration/7
78 FF 50 - 90 idle_inject/9
79 FF 99 - 139 migration/9
84 FF 50 - 90 idle_inject/11
85 FF 99 - 139 migration/11
104 FF 50 - 90 irq/9-acpi
108 TS - -20 39 kworker/R-ata_sff
112 FF 50 - 90 watchdogd
115 FF 50 - 90 irq/25-AMD-Vi0-Evt
116 FF 50 - 90 irq/26-AMD-Vi0-PPR
117 FF 50 - 90 irq/27-AMD-Vi0-GA
428 TS - -20 39 kworker/R-amdgpu_dm_hpd_rx_offload_wq
430 FF 50 - 90 card1-crtc0
431 FF 50 - 90 card1-crtc1
432 FF 50 - 90 card1-crtc2
433 FF 50 - 90 card1-crtc3
461 TS - 0 19 nvidia-modeset/deferred_close_kthread_q
462 FF 50 - 90 irq/62-nvidia
475 TS - 0 19 UVM deferred release queue
624 FF 50 - 90 irq/63-ELAN06FA:00
770 FF 1 - 41 psimon
842 FF 1 - 41 psimon
1255 FF 50 - 90 irq/96-rtw89_pci
1513 RR 99 - 139 rtkit-daemon
1589 FF 1 - 41 psimon
3098 RR 1 - 41 kwin_wayland
3098 RR 1 - 41 eDP-1
3098 RR 1 - 41 libinput-connec
3114 RR 20 - 60 data-loop.0
3115 RR 20 - 60 data-loop.0
3316 RR 20 - 60 data-loop.0
3316 RR 20 - 60 data-loop.0
3316 TS - 0 19 bazaarrunner
3316 TS - 0 19 krunner_charrun
3327 RR 20 - 60 data-loop.0
6596 TS - 0 19 WRRende~ckend#1
6596 TS - 0 19 WRRende~kend#19
37282 FF 1 - 41 psimon
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
37326 FF 80 - 120 cyclictest
sudo cyclictest --mlockall --smp --priority=80 --interval=200 --breaktrace=800
# /dev/cpu_dma_latency set to 0us
policy: fifo: loadavg: 11.16 5.94 3.72 19/2368 37630
T: 0 (37327) P:80 I:200 C: 268947 Min: 8 Act: 13 Avg: 13 Max: 462
T: 1 (37329) P:80 I:700 C: 76837 Min: 8 Act: 15 Avg: 13 Max: 78
T: 2 (37330) P:80 I:1200 C: 44819 Min: 9 Act: 15 Avg: 14 Max: 49
T: 3 (37331) P:80 I:1700 C: 31635 Min: 8 Act: 14 Avg: 14 Max: 153
T: 4 (37332) P:80 I:2200 C: 24443 Min: 9 Act: 15 Avg: 14 Max: 106
T: 5 (37333) P:80 I:2700 C: 19915 Min: 9 Act: 14 Avg: 14 Max: 132
T: 6 (37334) P:80 I:3200 C: 16802 Min: 9 Act: 19 Avg: 14 Max: 766
T: 7 (37335) P:80 I:3700 C: 14531 Min: 9 Act: 15 Avg: 14 Max: 179
T: 8 (37336) P:80 I:4200 C: 12800 Min: 8 Act: 21 Avg: 13 Max: 37
T: 9 (37337) P:80 I:4700 C: 11437 Min: 9 Act: 19 Avg: 14 Max: 264
T:10 (37338) P:80 I:5200 C: 10337 Min: 9 Act: 17 Avg: 14 Max: 73
T:11 (37339) P:80 I:5700 C: 9430 Min: 9 Act: 18 Avg: 14 Max: 83
# Thread Ids: 37327 37329 37330 37331 37332 37333 37334 37335 37336 37337 37338 37339
# Break thread: 37332
# Break value: 2488
sudo perf sched timehist > perf_timehist.txt awk '$1 >= 9980.792239 && $1 <= 9980.797096' perf_timehist.txt > perf_extract.txt https://pastebin.com/iuB98bXg
sudo perf sched latency > perf_latency.txt https://pastebin.com/JP0PfrSU
~~After digging a bit it looks like pipewire and spotify don't get realtime shedules only SCHED_OTHER and SCHED_BATCH~~ https://bbs.archlinux.org/viewtopic.php?pid=2279287#p2279287
4873 TS - 0 19 WRRende~kend#18
ludrol@bazzite:/var/home/ludrol/tmp$ ps -eo pid,class,rtprio,ni,pri,comm -L | grep -i spotify
8613 TS - 0 19 spotify
8613 TS - 0 19 spotify
8613 TS - 0 19 spotify
8620 TS - 0 19 spotify
8620 TS - 0 19 spotify
8620 TS - 0 19 spotify
8624 TS - 0 19 spotify
8625 TS - 0 19 spotify
8664 TS - 0 19 spotify
8664 TS - 0 19 spotify:cs0
8664 B 0 19 0 spotify:disk$0
8664 TS - 0 19 spotify:sh0
8664 TS - 0 19 spotify:gdrv0
8664 TS - 0 19 spotify:gl0
8664 TS - 0 19 spotify:sh1
8664 TS - 0 19 spotify:sh2
8664 TS - 0 19 spotify:sh3
8664 TS - 0 19 spotify:sh4
8664 TS - 0 19 spotify:sh5
8664 TS - 0 19 spotify:sh6
8664 TS - 0 19 spotify:sh7
8664 TS - 0 19 spotify:sh8
8664 B 0 19 0 spotify:disk$1
8664 B 0 19 0 spotify:disk$2
8664 B 0 19 0 spotify:disk$3
8664 B 0 19 0 spotify:disk$0
8664 TS - 0 19 spotify:sh0
8665 TS - 0 19 spotify
8669 TS - 0 19 spotify
8730 TS - 0 19 spotify
8804 TS - 0 19 spotify
8860 TS - 0 19 spotify
ludrol@bazzite:/var/home/ludrol/tmp$ ps -eo pid,class,rtprio,ni,pri,comm -L | grep -i pipe
2080 TS - 0 19 IpcPipeBroadcas
2080 TS - 0 19 IpcPipeBroadcas
2080 TS - 0 19 IpcPipeBroadcas
2080 TS - 0 19 IpcPipeBroadcas
3090 TS - -11 30 pipewire
3296 TS - -11 30 pipewire-pulse
3822 TS - 0 19 .NET EventPipe
3822 TS - 0 19 .NET DebugPipe
8613 TS - 0 19 pipewire-loop
40140 TS - 0 19 mpv/ao/pipewire
systemctl status rtkit-daemon
● rtkit-daemon.service - RealtimeKit Scheduling Policy Service
Loaded: loaded (/usr/lib/systemd/system/rtkit-daemon.service; enabled; preset: enabled)
Drop-In: /usr/lib/systemd/system/service.d
└─10-timeout-abort.conf
Active: active (running) since Thu 2026-07-16 08:42:27 CEST; 3h 3min ago
Invocation: 71422e850ec74c97b6edfa1666d3243a
Main PID: 1599 (rtkit-daemon)
Tasks: 3 (limit: 16290)
Memory: 580K (peak: 3.5M, swap: 12K, swap peak: 12K)
CPU: 92ms
CGroup: /system.slice/rtkit-daemon.service
└─1599 /usr/libexec/rtkit-daemon
lip 16 08:42:38 bazzite rtkit-daemon[1599]: Successfully made thread 3090 of process 3090 (/usr/bin/pipewire) owned by '1000' high priority at nice level -11.
lip 16 08:42:38 bazzite rtkit-daemon[1599]: Successfully made thread 3095 of process 3090 (/usr/bin/pipewire) owned by '1000' RT at priority 20.
lip 16 08:42:38 bazzite rtkit-daemon[1599]: Successfully made thread 3091 of process 3091 (/usr/bin/wireplumber) owned by '1000' high priority at nice level -11.
lip 16 08:42:38 bazzite rtkit-daemon[1599]: Successfully made thread 3103 of process 3091 (/usr/bin/wireplumber) owned by '1000' RT at priority 20.
lip 16 08:42:39 bazzite rtkit-daemon[1599]: Successfully made thread 3296 of process 3296 (/usr/bin/pipewire) owned by '1000' high priority at nice level -11.
lip 16 08:42:39 bazzite rtkit-daemon[1599]: Successfully made thread 3300 of process 3296 (/usr/bin/pipewire) owned by '1000' RT at priority 20.
lip 16 08:42:41 bazzite rtkit-daemon[1599]: Successfully made thread 3837 of process 3287 (/usr/bin/plasmashell) owned by '1000' RT at priority 20.
lip 16 08:42:42 bazzite rtkit-daemon[1599]: Successfully made thread 4081 of process 3287 (/usr/bin/plasmashell) owned by '1000' RT at priority 20.
lip 16 08:42:43 bazzite rtkit-daemon[1599]: Successfully made thread 4331 of process 4070 (/var/home/ludrol/.local/share/Steam/ubuntu12_32/steam) owned by '1000' high priority at nice level -10.
lip 16 08:42:44 bazzite rtkit-daemon[1599]: Successfully made thread 4332 of process 4070 (/var/home/ludrol/.local/share/Steam/ubuntu12_32/steam) owned by '1000' high priority at nice level -10.
I have my (1) project as homepage, if it doesn't load on browser startup then I have ze problem. (I am a monster with disabled tab restoring.)
IDK what I will do if I will have more than one, but that's future me problem.
Looks like since this video they changed the logo
On bazzite
Can't change my Plymouth boot screen
Can't set animated icons in toolbar (KDE)
The screen transition I chose is glitchy
The translation tries VERY hard to be different from windows (idk what's up with that as on other system it was normal(propably, I would need to make sure))
I don't know how to run containers from within containers, but that's a skill issue. (I wanted to build piefed from distrobox)
For the longest time I had to disable manually a module so my graphics tablet would work. It was fixed a month ago!
yt-dlp and Jdownloader2 work most times
Input 0000 and 0 are the same in the flow of your program.
Godot. You could try to see if Godot 3 will work better than Godot 4.
Godot 3 is a bit older but still is getting fixes and some new features. (long term support driven by community)
Googled it and this came out: https://musescore.org/en/node/14318