There has been some discussion of how build and configure an image, balancing tools, convenience, and performance. It raised questions like:
Does installing X on a system that will be used mostlly headless, cost much? Is it worth turning the service off by default?
I wrote a benchmarking tool to answer such questions. From the README.md
:
Benchmarking Audio System Performance
These files benchmark the audio processing capacity of a particular set up.
The benchmark runs a test to find the maximum number of test voices it can
play simulateously without getting buffer underruns (“xruns”). The test is
repeated at multiple buffer sizes (-p
injackd
).
You can find it here: