Re: avoiding VM page swaps for big sample buffers
Re: avoiding VM page swaps for big sample buffers
- Subject: Re: avoiding VM page swaps for big sample buffers
- From: David Duncan <email@hidden>
- Date: Sat, 25 Sep 2004 19:16:10 -0400
On Sep 25, 2004, at 06:52 PM, Frank (Steim) wrote:
Well, first of all a vector of 512 samples gives you a callback time
of a little more than 11 ms (the sample rate is 44kHz)
Yea, mental math mistake on my part. You would think I would realize
that 512*10 isn't nearly 44100 samples...
and second: yes in this vecotr I can make sample jumps (and
calculated crossfades and zero cross searches) which I want to be able
to do within my whole sample buffer.
Perhaps you should reserve your IOProc for simply copying data and do
all your calculations in a separate thread. You take the full latency
hit but simplify your callback. I'll leave it to you to determine if
your users would notice the latency.
Of course I could put a restriction in max buffer size and usually I
don't need such a big buffer (mostly just a few minutes long), but for
this partical project I need a really big buffer. (you can find a
demo of the program at: http://www.steim.org/steim/lisa.html)
Presumably your trying to get this working as quickly as you can, but
there are going to be things you need to change for Mac OS X. Huge
sparsely used buffers are one of those things that simply doesn't work
well on OS X.
--
Reality is what, when you stop believing in it, doesn't go away.
Failure is not an option. It is a privilege reserved for those who try.
David Duncan
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden