Am 14.06.2004 um 12:51 schrieb TLD:

> Suboptimal rules on the firewall?
> It happens to write rules that work but are slower than necessary, for
> example parsing UDP packets in the TCP+UDP rules instead of splitting 
> the
> execution paths into TCP rules and UDP rules.

Please reread my first post. I said pf on my firewall and ipf on my 
netbsd
box was disabled during the tests...

I got ~80Mbit/s by reducing the size of udp packets to 600 bytes (with 
ttcp)
it seems to be a better setting for my nfs setup also...

ciao,
schamil