[linux-audio-dev] [ANN] netjack-0.11

[torbenh at gmx.de]

On Mon, Apr 17, 2006 at 11:27:04AM +0200, Jens M Andreasen wrote:
> On Mon, 2006-04-17 at 01:05 -0400, Lee Revell wrote:
> > On Sun, 2006-04-16 at 23:22 +0200, torbenh at gmx.de wrote:
> > > 
> > > Work is underway in improving the latency for big channel counts,
> > > like 24in / 24out. There seems to be a bottleneck in the kernels
> > > handling of big UDP Packets. 
> > 
> > Nice, I'm looking forward to the LKML thread!
> > 
> > (you do plan to report it right...?)
> > 
> 
> Torben,
> 
> You are not checking the return value of sendto() ... It would be most
> interesting to know what these are when the application fails. EMSGSIZE
> EWOULDBLOCK and EAGAIN are some of the possible candidates.

i dont think its that interesting, return value of EAGAIN means, the
packet got dopped, and the sendqueue is full, this generally happens in
congestion situations at the dsl modem or something (where its never
reported...)

EMSGSIZE will not be set.

please note, that the application does NOT fail.

packet roundtrip time for big UDP packets is too high. So i cant run at
-l 1 and have to resort to -l 2 or even -l 3.

that means 3 periods of round trip latency.
meaning a remote fx processor would have 3 periods of latency added.



-- 
torben Hohn
http://galan.sourceforge.net -- The graphical Audio language