[linux-audio-user] Re: [linux-audio-dev] music engine

[Paul Davis]

On Mon, 2006-04-10 at 14:58 -0800, Patrick Stinson wrote:
> Paul, Everyone,
> 
> What kind of assumptions can we make about the performance of the
> various FIFO pipe/socket/whatever mechanisms on a POSIX machine? This
> is by no means my area of expertise, but I currently hold the
> assumption that a POSIX machine should provide FIFO pipe-ish things
> that are *very* fast, considering they dont live in a local context.
> If native FIFO I/O mechanisms are not fast enough for, say, spewing
> 1/256th of a 10-20 second wave form's data to some app, isn't there
> something wrong? What metrics do you keep on the tip of your tongue
> after working with ardour and jack?

they are definitely fast enough. the issue for me is the switch from
shared pointer access to data to having to fetch/read it via some IPC.
this is not a technical problem, its a "do i want to get my head around
this" issue. the answer is yes, but the answer to "when" is unclear.

ardour and JACK use in-memory FIFO's for interthread wakeups, and
although theoretically they are not RT safe, they are as good as it gets
unless we want to dump 2.4 kernels and use futexes.

the problem in ardour for waveform display is that it has to scale to
handle potentially enormous quantities of data. imagining zooming to
encompass all of a 4hr, 60 track session. although the number of pixels
on screen remains the same, the amount of data that represents is huge.
this forces careful threading design onto any naive attempt to "just ask
for the peaks and you're done".