From that I see you've got a collection of endpoints - several periodic ones. You will get at least 1 interrupt per transfer on a periodic endpoint, so if they were all scheduled in different microframes you would get 1k + 1k + 1k + (trivial numbers for 16ms and 32ms) interrupts per second. Bulk endpoints will interrupt when the queued buffer (which can be much larger than 512B) completes.
What does "vmstat 1" output before and after you start up the audio device? You can leave it running while the workload changes.
What does "vmstat 1" output before and after you start up the audio device? You can leave it running while the workload changes.
Statistics: Posted by jdb — Thu Jan 18, 2024 1:27 pm
