[07:20] <ForeverNoob[m]> Hello room
[09:18] <libera38> I have noticed that my Focusrite 2i2 (USB) has a lot more xruns using later kernels compared to 5.15. What have changed in the kernel since then that might impact?
[09:18] <libera38> Focusrite 2i2 Gen2
[09:19] <libera38> If I use 64 samples and 2 buffers using linux 5.19 I get extreme/consant xruns but using 5.15 I only experience every now and then. I need to use 64 samples and 3 buffers to at least get it to not sound like it is totally defect under 5.19
[10:01] <ForeverNoob[m]> When I hear stuff like this I always wonder if the scheduler has something to do with it.
[10:02] <libera38> When it happens it is quite often happening during updates on the screen. Not the extreme case though with 64 samples and 2 buffers that sounded extremely broken.
[10:03] <libera38> Using PekWM and no compositor has lower the xruns the most. That doesn't look too good though since PekWM doesn't update the parts screen that windows have been on and are no longer.
[10:19] <libera38> I posted about this on in the subreddit linuxaudio: https://www.reddit.com/r/linuxaudio/comments/yi7c09/recent_kernels_at_least_from_518_and_onward_gives/
[14:20] <Eickmeyer[m]> ForeverNoob: Per this https://wiki.linuxaudio.org/wiki/list_of_jack_frame_period_settings_ideal_for_usb_interface 2 buffers per period on a USB device is almost universally a bad idea, so that's usually the culpret.