[RDD] Latency and xruns [WAS: RD 2.0]

Fred Gleason fredg at paravelsystems.com
Sun Jul 11 10:31:29 EDT 2010

On Sunday 11 July 2010 02:12:17 am Aaron wrote:
> I'm just wondering if the audio engine can take that
> latency into account, and once audio is playing, start all the following
> sequenced audio events xxxx ms early to compensate.

It's possible in theory, but difficult in practice, especially with software 
that is designed to run across a wide range of hardware.  These latencies 
will vary according to lots of different parameters, including CPU type, 
memory, sound driver infrastructure, etc, etc.

The milling machine example that Cowboy used earlier in this thread is an 
illuminating one.  One of the reasons that design is able to achieve close 
to 'zero' latency is because it's *embedded* -- the target hardware is 
precisely known when developing the software and is completely dedicated to 
the task of running just that one machine (both important requirements when 
designing hard RT systems).  Neither of these things is true in the case of 


| Frederick F. Gleason, Jr. |               Chief Developer               |
|                           |               Paravel Systems               |
|  I know the answer!  The answer lies within the heart of all mankind!   |
|  The answer is twelve?  I think I'm in the wrong building.              |
|                                      -- Charles Schulz                  |

More information about the Rivendell-dev mailing list