Re: [ecasound] -ei limits

From: Kai Vehmanen <kvehmanen@email-addr-hidden>
Date: Sat Mar 05 2005 - 07:35:21 EET

On Wed, 9 Feb 2005, Eric Dantan Rzewnicki wrote:

> A little while ago you asked if there were any issues with resampling.
> The -ei effect is still limited. I've opened up my version in
> double highlimit = 25000.0f;
> I haven't done extensive quality tests, but it at least works. I
> couldn't open it up all the way to 256x as supported by libsamplerate.
> Doing so caused errors that looked like the engine was starving or
> something. Sorry not to be more specific. If I can find time I'll try to
> make a better report, but I just wanted to send a quick note to see if
> you might have time to look into it before releasing 2.3.6

I've now removed the upper limit altogether in CVS-tree. The original
reason for these arbitrary limits was to avoid segfaults in all scenarios.
I knew that there were some corner cases in the resampling code when
used with extreme values. And as these parameters can change in
unexpected ways (especially when controlled by GUI sliders/buttons,
oscillators, MIDI-sources), imposing the hard-limits was the easiest way
to protect against crashes (aside finding all the bugs of course, but as
always, time to debug is limited ;)).

The lower (hard-)limit is still there (max 8x slowdown). Ecasound also
still recommends an upper bound via the parameter hints (100x speed-up),
but the limit is a soft-one and you can force -ei to use even higher
values. I've tried to run various stress tests with dynamic control of the
resampling rate, and - after some bugfixes - I cannot produce any
segfaults anymore (tested both libsamplerate and internal resampler). So
this is hopefully safe for 2.4.0.

 Audio software for Linux!
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
Ecasound-list mailing list
Received on Sat Mar 5 08:15:09 2005

This archive was generated by hypermail 2.1.8 : Sat Mar 05 2005 - 08:15:09 EET