Re: [Rosegarden-user] MIDI Timing - Root vs. User

[Chris C. <ca...@al...>]

On Sunday 29 December 2002 19:00, Mark Knecht wrote:
> My thought about Jack effecting this is that since jack is granted
> root level capabilities by the kernel, Jack completely controls
> when things happen.

I think it's not just that JACK is given particular capabilities, but=20
that it uses those capabilities to run in a particular scheduling=20
mode that essentially means everything else is scheduled around it on=20
the assumption that it will only take very short slices of CPU time=20
at very regular intervals.  If it turns out to block, everything can=20
become absolutely fucked.  But I don't really know.

I would expect that if you run it as a normal user, JACK's performance=20
will not be as good but it will be incapable of affecting the=20
performance of anything else in this way.  It may well be worth=20
testing your setup again with JACK and its clients all run as=20
non-root users, because from my very limited perspective I think that=20
even with non-root users you should still get better performance than=20
the example file you've given, and it would eliminate one possible=20
variable when trying to work out what's going wrong.

> I'm wondering whether Jack having this level of access to machine
> resources will get in the way of RG sending and receiving MIDI
> events?

Yes, absolutely, but it'd only make a serious difference if JACK went=20
really badly awry.


Chris