Package: mpd Version: 0.14.2-2 Severity: normal Between version 0.13.2-3lenny1 and 0.14.2-2, mpd has gotten much worse at mp3[1] sound playback on armel. It seems to be using a *lot* more cpu.
I noticed that when I'm running the stable version, mpd's cpu usage on my thecus while playing a mp3 is often below 1%. With the unstable version, it never gets below 20-30%. This isn't just about CPU load numbers of course. It's about reliably getting clear playback under load without audio glitches caused by the buffer running dry, etc. I can reproduce this at will, on two separate arm machines. They are: turtle: a thecus N2100, with this: Bus 002 Device 002: ID 041e:3040 Creative Technology, Ltd SoundBlaster Live! 24-bit External SB0490 box: a qnap ts-109 with the following cheap usb sound dongle: Bus 002 Device 005: ID 0d8c:000c C-Media Electronics, Inc. Audio Adapter My first guess is that something has been rewritten and is using floating point, at which of course, arm sucks[2]. Grepping for "float" in the new source tree certianly turns up a dozen more than in the old. pcm_resample_libsamplerate.c looked particularly likely, at a guess. But configuring --disable-lsr didn't help. Nor did trying all the samplerate_converter settings, much. I'm still learning toward it being due to floating point, but since mpd has become threaded, arm threading innefficies also seem plausible. -- see shy jo [1] I think ogg is also affected, but I've been testing with mp3s only. [2] armel sucks less, but let's face it: w/o cpu-specific optimised builds, it still sucks :-)
signature.asc
Description: Digital signature