12-16-2018, 08:42 PM
(This post was last modified: 12-17-2018, 08:05 AM by mactolinux.)
Funny thing: I did many tests (looong tests) with i2s recording to SSD/HDD, RPi as a slave, in real-time modes (at high rates, 192-24/32, with SOX, arecord, Jack, whatever), and discovered that real-time tuning is something that put linux (RPi are not power heroes) into a very stressed situation... As result – recording has lot of mistakes. Instead, regular, almost out-of-the-box Lite Raspbian with basically generic multi-tasking setup is exact in what is it recording (bit to bit, literally). With some very light config tuning.
How is this linked to a regular playback procedures? I don't know, but I would take RT/low latency tuning with care, again comparing before and after, thanks it is easy to do by changing SD-cards.
P.S. You may find sometime even a bit curious stories, like "why am I getting overruns during recording? – you shall go to RT settings and use jackd, no overruns there". Almost true, but... when you look inside the data flow and compare it with the source, you see more problems than before...
How is this linked to a regular playback procedures? I don't know, but I would take RT/low latency tuning with care, again comparing before and after, thanks it is easy to do by changing SD-cards.
P.S. You may find sometime even a bit curious stories, like "why am I getting overruns during recording? – you shall go to RT settings and use jackd, no overruns there". Almost true, but... when you look inside the data flow and compare it with the source, you see more problems than before...
