Re: [Jack-Devel] Testing JACK and PA latency

On Thu, September 19, 2013 1:59 am, Paul Davis wrote:
> you're trying to measure the time between A and B where you have access to
> A (or specifically, a way to carry out measurements there) but no access
> to
> B.
>
> i don't see how this can work. it is a bit like asking "how long does it
> take to get from paris to brussels" but when you are unable to cross the
> border between france and belgium, and cannot talk to anyone in brussels.
>

Yes, exactly. It's a conundrum :-)

Can anyone suggest an app that will provide a round trip back into jack
via pulse sink/source?

Perhaps I need to write something specifically for that task that will
communicate with jack and pulse at the same time? Or at least be able to
retrieve useful information...






>
> On Wed, Sep 18, 2013 at 11:29 AM, Patrick Shirkey <
> [hidden]> wrote:
>
>>
>> On Wed, September 18, 2013 11:36 pm, Paul Davis wrote:
>> > there isn't any way to separately compute the "outbound" and "inbound"
>> > latency. in fact, jack_iodelay's message about the value to use for -I
>> and
>> > -O is based on the not-entirely-defensible but somewhat reasonable
>> > assumption that the outbound and inbound latencies are symmetric.
>> >
>> > put differently, i don't think there is any way to do what you want.
>> > someone may prove me wrong.
>> >
>>
>> It's a new one for me too so I'm not expecting a pre-existing solution.
>>
>> This is part of a wider test procedure so I am trying to get a method
>> that
>> can be considered robust and literal. I'm sure it is viable to calculate
>> the latency for each step in the process based on configuration settings
>> but in this case it will be *very* useful (and potentially lucrative for
>> everyone) to present a definitive result and test procedure.
>>
>> Any ideas for how to go about it are welcome.
>>
>> To recap I am seeking to define the i/o latency at any period
>> size/sample
>> rate when an application that is dependant on PA is sending/receiving
>> data
>> via JACK.
>>
>> If anyone has thoughts on a method for tweaking the configuration
>> settings
>> for both PA and JACK while running (automated) latency tests that would
>> also be very useful. Piece of cake right?
>>
>>
>>
>> >
>> > On Wed, Sep 18, 2013 at 6:55 AM, Patrick Shirkey
>> > <[hidden]
>> >> wrote:
>> >
>> >>
>> >> On Wed, September 18, 2013 4:49 pm, Patrick Shirkey wrote:
>> >> > Hi,
>> >> >
>> >> > Can anyone recommend a rigorous procedure for verifying round trip
>> >> latency
>> >> > between i/o using a combination of both JACK and PA?
>> >> >
>> >> >
>> >>
>> >> I have the following setup:
>> >>
>> >> jack running in duplex mode (1024/48000/2)
>> >> pa connected to jack with sink/source plugins
>> >> skype* running via pa
>> >> jack_iodelay
>> >>
>> >> I have the following connections
>> >>
>> >>      jack_delay (out) -> pa jack source (1)
>> >>      system_in (1) -> jack_delay (in)
>> >>      system_in (1) -> pa jack source (1)
>> >>      pa jack sink -> system_out (1)
>> >>
>> >> I can hear the output signal from jack_delay in the skype call test
>> >> service.
>> >> I can hear the mic input in skype call test service
>> >> I can hear skype output in my speakers via jack system output
>> >>
>> >> I see the following info in jack_delay console messages:
>> >>
>> >>  22120.239 frames    460.838 ms total roundtrip latency
>> >>         extra loopback latency: 21096 frames
>> >>         use 10548 for the backend arguments -I and -O ??
>> >>
>> >>
>> >> However I don't think this setup is calculating latency details that
>> I
>> >> am
>> >> looking for.
>> >>
>> >> I would like to retrieve the following info:
>> >>
>> >> 1: latency from jack system in -> pa_source -> skype in -> skype
>> network
>> >> output (before the data is transmitted to the external receiver)
>> >>
>> >> 2: latency from skype network input (immediately data that is
>> received
>> >> from external provider) -> pa_sink -> jack_system out
>> >>
>> >>
>> >> * skype is used for testing i/o via PA. Any other duplex app could
>> also
>> >> be
>> >> used.
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> --
>> >> Patrick Shirkey
>> >> Boost Hardware Ltd
>> >> 
>> >> Jack-Devel mailing list
>> >> [hidden]
>> >> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
>> >>
>> >
>>
>>
>> --
>> Patrick Shirkey
>> Boost Hardware Ltd
>> 
>> Jack-Devel mailing list
>> [hidden]
>> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
>>
>


--
Patrick Shirkey
Boost Hardware Ltd

1379572490.19746_0.ltw:2, <36869.86.127.135.98.1379572479.squirrel at boosthardware dot com>