> I haven't measured it - but think of outgoing packets being fragmented =
by
> IP - they're back-to-back using - obviously - the same destination
> address.

I'm not sure exactly how it all works, but I'd assume the performance wou=
ld=20
have to do with how long it would take to do a lookup in the routing tabl=
e,=20
and therefore be determined somewhat by the size of the routing table.