[15:09] <Eickmeyer> OvenWerks: I got an email from Andrea Righi at Canonical. I'm going to include you on the conversation, but they're talking about merging the lowlatency features of the lowlatency kernel into the generic kernel and making the generic kernel basically better-than-lowlatency.
[15:11] <Eickmeyer> They want to know how to best benchmark it. I'm thinking running JACK with as low of a buffer as possible and watching for xruns.
[15:12] <Eickmeyer> Also watching the DSP.
[15:30] <OvenWerks> Eickmeyer: I am not sure if I will take my laptop or not but I don't think we will have inet anyway.
[16:07] <Eickmeyer> Going on a trip OvenWerks ?
[16:08] <Eickmeyer> Basically just was looking for your input.
[16:49] <arraybolt3> Eickmeyer: I'm not sure if my input is any good, but you might try using some MIDI files with something like LMMS or probably some other MIDI player, run them through some plugins and whatnot, and try to make a CPU-intensive audio production simulator that gets as close as possible to maxing out a system without causing xruns during low-latency usage.
[16:49] <arraybolt3> Once you have something that gets 0 xruns but only just barely, then switch kernels and run the simulation again.
[16:51] <arraybolt3> (My systems aren't all that powerful so it's not hard to max them out :P I think one LMMS instance and one Rakkarack instance (is that how you spell that?) is enough to get xruns on my Z220 machine.)
[16:52] <arraybolt3> Not to mention the fact that it sounds pretty horrible...
[20:25] <Eickmeyer> arraybolt3: I'm just going to have them to the 24-hour test.
[20:26] <Eickmeyer> Have the system idle with the default jack/pulseaudio bridge setup as low of a buffer as possible without xruns, get the average DSP, and let it run overnight and record how many Xruns after 24 hours.
[20:26] <Eickmeyer> Do that for the experimental kernel, the generic kernel now, and the lowlatency kernel now.