> On 29 Nov 2015, at 12:57, James Harkins <[hidden email]> wrote:
> I'm not looking for 100% exact sample accuracy. > However, the timing of OffsetOut should be a noticeable improvement over
Out, and it simply isn't.
> If it is no improvement, then it's cruft to be removed. I can neither confirm your test result nor does it confirm my other tests (both here on OSX) Instead of f.reject(_ == 441) -> [ 482, 439, 484, 439, 484, 449, 439, 483, 438 ] I get f.reject(_ == 441) -> [ 477 ] So these 10 seconds include one of the typical realtime adjustment deviations. As just posted on users - I reproduced timing tests with OffsetOut by Gerhard Eckel on OSX: BTW I forgot a useless bit of code in there which inihibits recheck: in SynthDef \diracs drop '* EnvGate(fadeTime: 0)' What's not posted in this cited thread - but what I did also - was testing with Out instead of OffsetOut, and OffsetOut gave a clear improvement of timing. What you describe seems to be related to Linux or/and audio setup. Regards Daniel |