[LAU] JACK MIDI latency -- was Re: Concerning schedtool

[Robin Gareus]

On 04/20/2014 03:49 AM, Ralf Mardorf wrote:
> IIUC we are not talking about the same alsarawmidi.
> 
> On Sat, 2014-04-19 at 18:18 +0200, Ralf Mardorf wrote:
>> > jackd --sync -Xalsarawmidi -dalsa -r$sample_rate -p$frames_period
>  ^^^^^        ^^^^^^^^^^^^^
> On Sat, 2014-04-19 at 19:47 +0200, Robin Gareus wrote:
>> > jackd -S -dalsa -dhw:2 -r48000 -p64 -n2 -Xraw
           ^^^^^^                         ^^^^^
it prints
JackALSARawMidiDriver::Start - Starting 'alsarawmidi' driver.
JackALSARawMidiDriver::Start - starting ALSA thread ...
JackALSARawMidiDriver::Start - started ALSA thread.
as part of the startup message.

-Xraw is needed because otherwise I don't get any midi ports at all.
fPlaybackChannels and fCaptureChannels in
<jack2-source>/linux/alsarawmidi/JackALSARawMidiDriver.cpp are zero
otherwise.


Devin wrote:
> when I requested MIDI latency/jitter test results for
> the driver, I saw peak jitter results as low as 60 microseconds.

The jitter measurement here using alsarawmidi is 100 us stddev over 10K
events, 3 bytes each (using a USB2 device - details in prev email).

That's not far off from what Devin quoted, but without knowing what and
how he measured it, I can't comment on it.

ciao,
robin