[LAU] 96 kHz -- a bottleneck somewhere

[Fons Adriaensen]

On Sat, Apr 26, 2014 at 05:23:07PM +0000, Jonathan E Brickman wrote:

> 1. At 96 kHz, schedtool definitely matters. 
> Taking it out increased xruns a lot. 

That means that the non-real-time part of your app is
somehow able to modify the scheduling of the real-time part.
In other words a design problem. If the interface between the
two parts is OK you should be able to completely block the
non-real-time part without any effect on the audio or xruns.

> 2. Some of my patches -- especially strings-related -- are
> quite a lot improved by the shift to 96 kHz, the audio detail
> at the high third of my 88 is much better.

That again indicates a design problem, this time with the
synthesis algorithms.

> Not very surprising from a mathematical point of view of course.

You must be using some very weird mathematics then. 

You can of course continue to claim things like this, and
that you get less latency using zita-ajbridge (which is 
impossible), but you'll be jumping the shark quite soon.


-- 
FA

A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)