Neue MoOde Version 3.8:
MoOde ist ein stabiles Audio-OS für die raspberri Pis. Es kostet mittlerweile 10 Dollar, wird von Tim Curtis aus Detroit sehr gewissenhaft und fachkundig weiterentwickelt und zählt zu den besten Betriebssystemen, die ich für Audio-Anwendungen auf dem raspi kenne. Es bietet u.a. die Möglichkeit, verschiedene Kernel auszuwählen, ebenso die CPU-Prioritäten und die Art und Weise, in der der MPD die Tasks abarbeitet, wenn ich es richtig verstanden habe. Tim schreibt dazu:
"Hi Luki,
I can offer the technical information below which is a reprint of the (i) help in System config screen, plus some additional info.
Linux kernel
============
Standard: mainline kernel for Raspbian.
Advanced-LL: low latency advanced audio kernel.
Advanced-RT: real-time advanced audio kernel.
LL and RT kernels support advanced I2S audio capabilities.
- Up to 384k sample rate for ES9023 and PCM512x codecs.
- Integer bclk divider for ES9023 codec on 48k/96k rates.
- DSD128 format support via 352.8k rate.
- Bypass OS filter on PCM512x, PCM514x, PCM524x via 384k rate.
- Fixed 64fs bclk ratio for 2 channel, 32/24/16 bit audio.
DISCLAIMER: Advanced kernels are experimental without any support or warranty.
Install can take several minutes. Reboot after completion message appears.
Additionally, the Advanced kernels do two things. One is that they include some audio capabilities that are not available in mainline Linux kernel although 384K rate is now in Linux mainline as of 4.9.40
Two, they are configured and compiled with settings that enable a Low Latency or Real-Time task scheduling system. Tasks for example MPD or Squeezelite can be priortized so they are less likely to be preempted (yield the CPU to other tasks). Many computer audio enthusiasts perfer Low Latency or Real-Time kernels.
CPU Governor
============
On-demand: dynamically scales cpu frequency based on demand.
Performance: runs the cpu at maximum frequency.
Additionally, the Performance setting, since it runs the CPU at max frequency, will provide a more responsive moOde UI and perform tasks more quickly.
MPD scheduler policy
====================
TS: standard round-robin time-sharing policy.
FIFO: real-time fixed-priority First-In First-Out scheduling policy.
RR: real-time fixed-priority Round Robin scheduling policy.
Additionally, a task that is assigned the FIFO policy only yields the CPU when it is preeempted by a higher static priority task or if it voluntarily yields the CPU for example while waiting for an I/O to complete. The FIFO policy makes it less likely that the task will be preempted but can result in the task hogging the CPU and blocking other tasks.
A task that is assigned the RR scheduler policy runs for a certain time quantum and then yields the CPU. Within the time quantum that it runs it behaves exactly like FIFO. This reduces the likelyhood that the task will hog the CPU.
A task that is assigned the TS scheduler policy is set to static priority 0, the lowest priority and thus it can be preempted by FIFO and RR tasks at higher static priorities. "
Ich verwende es mit einem Raspi3 und einem Allo Boss DAC. Bin recht zufrieden. Insbesondere die Möglichkeit, den internen Filter durch upsampeln per SoX auf 384k zu umgehen, hört sich überraschend gut an.
Luki