[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Please Improve Debian for Multimedia Production

Cassiel wrote:

2009/3/26 Tzafrir Cohen <tzafrir@cohens.org.il <mailto:tzafrir@cohens.org.il>>

    I'm trying to follow this thread, but I'm not sure everybody here even
    using the same terminology.

    On Thu, Mar 26, 2009 at 02:01:22PM +0100, Cassiel wrote:
    > Hi
    > 2009/3/26 Adrian Knoth <adi@drcomp.erfurt.thur.de
    > > On Wed, Mar 25, 2009 at 09:05:32PM +0100, Grammostola Rosea wrote:
    > >
    > > > The problem is an realtime kernel and a proper configuration
    for music
    > > > production. Without that, you better stay away from music
    > >
    > > Not really. I absolutely have no problems with CONFIG_PREEMPT,
    > > ondemand scheduler and all this fancy stuff in a vanilla
    2.6.29 running
    > > on a dualcore amd64 laptop.
    > >
    > > It's an outdated myth that you need RT kernels for audio
    > > though it helps with very low latencies. (let's say buffer
    sizes below
    > > 10ms and high CPU loads)
    > >
    > >
    > maybe it is worthwhile to clarify where "audio production"
    begins and "audio
    > fun" ends.
    > Recording 16 tracks + monitoring could be an impossible task
    with your !=RT
    > kernel.

    Could you please be more specific?

    What extra patches do you think need applying?

    What changes to the configuration are needed?

    Vs. what upstream kernel version?

    References, supporting benchmarks and such would be an added bonus.

When talking about RT kernel only one patch comes into play -->> http://www.kernel.org/pub/linux/kernel/projects/rt/

other references regarding kernel config are found in -->> http://www.alsa-project.org/main/index.php/Low_latency_howto

IMHO, the upstream kernel version should follow the RT patch releases.
That would be a nice thing imho, yes.

Benchmarks can be easily provided by users if encouraged to use the same toolbox, eg.-->> http://git.kernel.org/?p=linux/kernel/git/tglx/rt-tests.git;a=summary Actually I have none but as I introduced in the previous post recording one more track on my i386 PentiumIV dual core running Ardour with 14-16 tracks + fx + software monitoring (i.e. ardour) is possible running RT kernels, running stock kernels results in too many xruns.

Much info about RT kernels you find here:


Reply to: