["Lars Ruoff" <lars.ruoff@xxxxxxxxxxxxxxxxxx>]

> > (However CPU consumption is negligible
>
> ...but only if you don't use a filter for the RTP streams tap.
>
> If you do, it's not as fast as you might think when reading in the file.
>
> Just passing NULL rather than (a copy of) "rtp && ip && !icmp") in the
> "register_tap_listener()" call in "register_tap_listener_rtp_stream()"
> reduced the user-mode CPU time required to read in one large capture I
> have from 68.5 seconds to 28.6 seconds.

woah! ok. when i was doing the test with and without the RTP stream tap
registered on startup, the differences in redissection where within
measuring inaccuracies.
?!
Anyway, i will try to get rid of those filters

Lars