You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Ideally we should be able to have a single buffer queue, with all the other limits disabled on the tee pads. This would ensure minimal latency for the software volume changes. And in general having low latency in the pipeline is a good thing.
However, if you try and implement this as things are now it will fail. But not consistently, appsrc based backends like spotify seem to be better at triggering the lockups. After a lot of debugging the culprit has been pinpointed as the fakesink with sync set to true. This element was added per the recommendation in https://coaxion.net/blog/2014/01/gstreamer-dynamic-pipelines/ to ensure that when all dynamic outputs get disabled we still playback using the clock, and don't just speed through everything.
Currently I don't know how we can solve this. So for now this is just a quick summary / brain dump so we have the problem documented.
The text was updated successfully, but these errors were encountered:
Ideally we should be able to have a single buffer queue, with all the other limits disabled on the tee pads. This would ensure minimal latency for the software volume changes. And in general having low latency in the pipeline is a good thing.
However, if you try and implement this as things are now it will fail. But not consistently, appsrc based backends like spotify seem to be better at triggering the lockups. After a lot of debugging the culprit has been pinpointed as the fakesink with sync set to true. This element was added per the recommendation in https://coaxion.net/blog/2014/01/gstreamer-dynamic-pipelines/ to ensure that when all dynamic outputs get disabled we still playback using the clock, and don't just speed through everything.
Currently I don't know how we can solve this. So for now this is just a quick summary / brain dump so we have the problem documented.
The text was updated successfully, but these errors were encountered: