Now we compare them with kernel profile stats:
af_packet_send: 7876 ms
nested dev_queue_xmit: 3714 ms
net_rx_action: 1464 ms
af_packet_recv: 411 ms
For send gprof tells 9930 ms and kernel 7876 so 2 sec in difference
is generic socket and system call handling overhead (and measure error).
The recv gproffed is 3540 ms, kernel 1464+411 so there is again about
2 sec overhead.
netflow simulator :) but the 2.4 on PII/375 can handle up to 30000 pps
in userspace loop.
Without userspace overhead and not bouncing packets (like in my test)
the router based on 2.4 could handle about 70000 pps without qos,routing
and filtering overhead ..
Someone tested in real world ?
best regards, devik
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/