Re: [Jack-Devel] Latencies...

On Fri, Sep 2, 2011 at 6:17 AM, Stéphane Letz <[hidden]> wrote:
> Hi All,
>
> I the process of looking at latencies stuff again, I saw that even on OSX (when latencies are somewhat reported by the CoreAudio driver itself...), we end up with differences between what is measured by the jack_iodelay tool, and what is reported by jack_lsp -l
>
> I came also across this paper: http://cluster.inf-ra.uni-jena.de/~adi/icimt10.pdf
>
> But more fundamentally, would il make sense to add an option in jack_iodelay tool to automatically "adjust" all physical I/O port latencies? (after the measurement has been made...)? So that a user would typically launch JACK, start  jack_iodelay with this special option and then I/O latencies would be correct?

this seems potentially useful, indeed. but ...

> Or we could even imagine a more integrated way of using jack_iodelay code to do the job at JACK server launch time?

we could imagine it, but since its use requires a rather odd/unusual
physical connection setup (to create a loopback), it seems that it
could not really be used by most people most of the time.

this applies equally to your first idea, which would require a
loopback on every physical port, which i know for me is totally
unrealistic (RME Digiface).

1314962464.9539_0.ltw:2,a <CAFa_cKm=wp_1RhFDuh4cy=knn8u1MDy7At-N4E_C3ky3G41kYQ at mail dot gmail dot com>