[ubuntu-studio-users] "zero-latency" - Was: Sorry for repost, this is the finalised reply. With all the wrong spelling / erata removed.

Mike Holstein mikeh789 at gmail.com
Thu Dec 5 18:32:21 UTC 2013

On Thu, Dec 5, 2013 at 1:28 PM, Ralf Mardorf <ralf.mardorf at alice-dsl.net>wrote:

> Hi :)
> I try to correct the fallacy about latency and to explain it in a
> non-exact, non-technically way.
> On Thu, 2013-12-05 at 17:41 +0000, Alex Armani wrote:
> > zero-latency kernel
> Even the RT patch with full RT enabled is even not hard real-time
> capable, let alone that zero-latency is possible in reality ;).
> There even is latency before the light from your display touches your
> eyes and before your perception is able to form a view of what is
> displayed. There is no reality with zero-latency.
> Now, important for audio engineering is to understand the differences
> between different kinds of preemption and preemption rt.
> Hard real-time e.g. is possible with stand alone gear and oldish
> computers, such as the C64, because for such gear and old computers
> there's direct access to the hardware, without any layers. For a PC this
> is impossible, even not with a kernel-rt. The lowlatency kernel is
> relatively far away from a kernel-rt. You will notice that humans are
> able to handle latency, since everything in our live comes with latency,
> for example analog music instruments have got latency. The issue caused
> by PCs is jitter and the low latency kernel does still produce much MIDI
> jitter. You can get rid of more, but nor all jitter, if you use full
> real-time capabilities enabled, by a patched kernel. Clean audio
> production sometimes can be done with a vanilla kernel, since audio
> jitter belongs much to the kind of audio card and a vanilla kernel on a
> fast PC sometimes does provide more than is enough to get less jitter,
> sure you can lower the latency, the harder the kernel does enforce
> preemption rt.
> The more layers or regarding to the kind of protocol a layer does talk
> to other layers, the higher the latency and the more jitter. I guess the
> preemption rt processes within the kernel does optimize priorities, but
> can't do much about issues caused by layers.
> At least there definitively is no zero-latency possible, even if PC and
> all the layers once should be more precise than old computers were.
> Regards,
> Ralf

i read that too, and took note, and assumed "zero-latency" was just a
quickly mis-typed reference to the "low-latency" kernel.. though, it is
nice to have factual clarification in case someone was wanting copper wire
like latency (which is not zero either) and started searching for "ubuntu
zero latency kernel"...

> --
> ubuntu-studio-users mailing list
> ubuntu-studio-users at lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-studio-users


mikeholstein.info <http://www.mikeholstein.info/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-studio-users/attachments/20131205/24ebb6e5/attachment-0001.html>

More information about the ubuntu-studio-users mailing list