[LAU] Estimating JACK latency

[Joel Roth]

On Tue, Apr 17, 2012 at 05:09:52PM +0000, Fons Adriaensen wrote:
> On Mon, Apr 16, 2012 at 05:21:47PM -0400, S. Massy wrote:
> 
> > Okay, so for a standard 2 periods of 128 frames each, my nominal dddelay
> > is 256 frames or 5.3ms at a sr of 48k, twice that for a round-trip.
> 
> No. The nominal roundtrip-delay (physical input to physical output
> if you connect the corresponding ports directly) is period * nperiods.
> 
> 1 * period occurs at the input, (nperiods - 1) * period at the output.

Thanks for clarifying this.

I understand that the round trip delay for the soundcard is the sum
of capture port latency and playback port latency, plus whatever
additional delay is associated with the AD/DA converters.

However I see what appears to be a discrepency for
the latencies reported by jackd (2 + 2 periods roundtrip)
and by jack_lsp -l (1 + 2 periods round trip).
This is for jackd2.

jackd says:

configuring for 44100Hz, period = 1024 frames (23.2 ms), buffer = 2 periods
ALSA: final selected sample format for capture: 32bit integer little-endian
ALSA: use 2 periods for capture
ALSA: final selected sample format for playback: 32bit integer little-endian
ALSA: use 2 periods for playback

However, jack_lsp -l says:

system:capture_1
        port latency = 1024 frames
        port playback latency = [ 0 0 ] frames
        port capture latency = [ 1024 1024 ] frames
system:capture_2
        port latency = 1024 frames
        port playback latency = [ 0 0 ] frames
        port capture latency = [ 1024 1024 ] frames
system:playback_1
        port latency = 2048 frames
        port playback latency = [ 2048 2048 ] frames
        port capture latency = [ 0 0 ] frames
system:playback_2
        port latency = 2048 frames
        port playback latency = [ 2048 2048 ] frames
        port capture latency = [ 0 0 ] frames

Is this inconistent, or is there some other explanation?

> If you masure it using jack_delay you will in most cases find a value
> that is a bit higher, this is due to delay in AD/DA converters etc. For
> some types of soundcard (e.g. USB) the real latency can be significantly
> higher than the nominal value. You can tell Jack about this using the 
> -I and -O options.
> 
> 
> The extra delay when making a loop such as Ardour -> jconvolver -> Ardour
> is one period.
 
Great to have these details available for unschooled
developers as myself!

Regards,

Joel
 
> Ciao,
> 
> -- 
> FA
-- 
Joel Roth