07-23-2019, 04:25 PM
(07-22-2019, 11:18 PM)Tim Curtis Wrote: Thats not good since Buster is the new platform. We should retest on moOde 6. I should probably do a beta release since 6 is based on the new Raspbian Lite 2019-07-10 release and 4.19.58 (or higher) kernel. There are tons of bug fixes in the new Raspbian release.
Raspbian Buster is evolving very rapidly :-)
Doh!
It's official. I'm suffering old-timers disease.
Remember that RPi4B stress test with SoX resampling cranked up to 32bits/384kHz? I forgot to reset it before I started playing with CD playback in moOde 5.4beta2. Since I simply transferred the uSD card when I repeated the test on an RPi3B+, the same setting applied on that SBC too.
Finally saw what I'd done when I cranked up the MPD log_level to verbose after I noticed MPD was throwing "Decoder is too slow..." messages.
Haven't noticed any dropouts with moOde 5.4beta2 on the RPi4B since I disabled SoX resampling.
OTOH, with moOde 5.3.1 on an RPi3B+, I don't seem to suffer dropouts with or without SoX resampling to 32bits/384kHz enabled, whereas I did with moOde 5.4beta2 and resampling enabled.
All this, by the way, with a 16bit/48kHz output stream through a Creative Technologies USB-BT adapter to my Sennheiser BT headphones.
On the whole, a happy result but it suggests something subtly different between Raspbian Stretch and Buster. (It also suggests I need better memory aids, but that's a color of a different horse.)
Things which might still be done: See how much resampling is possible without dropouts. See if multithreading the SoX resampling makes a difference. Candidly, I think this is just busy work. The root cause lies elsewhere.
Regards,
Kent
