Re: [linux-audio-dev] new apple platform viability

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] new apple platform viability
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Fri Jul 21 2000 - 18:02:00 EEST


>Hi Bill,

Big mouth here, butting in again:

>just a question related to Snd:
>is it possible to show the show the moving playback pointer during play ?

it is. you need to set a guile variable. its name eludes me right now.

>This would be VERY useful when locating audio sections.
>(Of course this has to work in synergy with the current audio buffersize,
>which means that you have to delay the visual playback pointer position
>by the buffer audio latency, or the visual representation will not be in
>sync)

BS. you cannot do anything with a GUI+keyboard+mouse+human even
remotely quick enough to make this matter. if you try to place a
"marker" just when you hear something, you will never get it to within
any reasonable buffersize resolution (unless you're running with a
huge buffersize). you have to plan on placing the marker approximately
and then moving it using other means.

>snd uses 45% (!!) of CPU on my PII400 when playing a simple stereo WAV
>(is this due to the reverb and resampling routines always active ?)

you must be doing something odd. i've never seen this.

>In my opinion it should consume much less CPU. Plus if I move or
>resize the window, I hear lots of dropouts on a totally idle machine.
>Do you use a threaded approach ? (the playback thread could be made
>SCHED_FIFO)

snd is not threaded. however, as its integration into ardour shows,
its not particularly hard to do this (ardour effectively runs snd as 2
threads). it changes snd's internal operation from a "push" model (as
in snd itself) to a "pull" model (another thread requesting data from
snd), but almost no code needs to be changed to make this work.

>The speed when resizing windows is very fast, how do you do the replotting.
>(eg do you keep a low res version of the wave in memory ?)

what you see is not the wave. its an amplitude envelope. bill's been
pretty smart about this.

snd is an awesome program. its problems for me aren't in the areas
that Benno is worrying about - these are easy to solve or are
non-issues in real-life. instead, my concerns have to do with the GUI,
which is extremely useful for an emacs user and seasoned snd user, and
very unhelpful to a novice. lots of incredible features are only
available by typing Guile/Scheme expressions. etc.

and believe me, the GUI in snd is by far the most complex part of the
program, at least to someone who is coming to look at the source for
the first time.

--p


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Fri Jul 21 2000 - 19:31:23 EEST