7 Jun
2021
7 Jun
'21
8:15 p.m.
On 6/7/21 5:29 PM, Fons Adriaensen wrote:
On Mon, Jun 07, 2021 at 05:11:57PM +0200, Robin Gareus
wrote:
In case of Ardour, the roundtrip is measured.
For USB, there is an additional queue inside the ALSA driver. If I
understood correctly its average fill level depends on when USB endpoint
is started relative to the ALSA device start. It relaates to playback
only (not capture).
You may remember the discussion from 11 years ago where Clemens
explained this:
https://web.archive.org/web/20151001012055/http://jack-audio.10948.n7.nabbl…
It seems the kernel patch mitigates this issue. Starting the playback
endpoint in sync with capture might reduce the randomness of the
driver's systemic latency.
--
robin
Does this help to reduce the random systemic
latency of USB/ALSA devices?
Currently with USB device there is up to the buffer size (but at most
20ms) additional I/O latency. The delay is different every time the
device is started, or after an xrun occurs. It would be great if this
doesn't need to be measured every time.
Is that roundtrip delay, or input/output separately ?