Hi David,
We meet again - I guess you're the only person answering questions in
the Linux community ;-)
The first interface, eth0, will show all kinds of performance issues.
The other ones, eth1, eth2, and eth3, should be mostly fine.
Actually, in this case I'm using eth0 as an admin interface, and
eth1-eth3 for the experiments.
There are some other anomalies, though. eth2 and eth3 (BCM5704) exhibit
slightly worse performance than eth1, though this may be attributed to
the different card or the bus (I think eth0 and eth1 are on-board).
Also, the interrupt rate varies considerably over time, from around
4000/sec to over 15000/sec, with no apparent reason. If I understand the
driver code correctly, the interface should be constantly polling, so
the interrupt rate should be fairly steady (and low).
None of this explains the duplex issues, though. Unfortunately, there
are too many variables involved, compared with the other machines I'm
experimenting with (processor type and speed, NIC make and model, etc.).
--Elad
--
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
