Now if we can cut that second column to all 0s we are set ;->
cheers,
jamal
On Thu, 4 Oct 2001, Robert Olsson wrote:
>
> Correct!
>
> We were interested to see the the packet distribution on the different CPU's
> with this scheme. Reality had data we couldn't get in the lab. Seems there
> is nothing to complain about. Yet. I'd better say. :-)
>
> cat /proc/net/softnet_stat
> 2b85c320 0000d374 6524ce48 00000000 00000000 00000000 00000000 00000000
> 0007119d
> 2b8b5e29 0000d615 653eba32 00000000 00000000 00000000 00000000 00000000
> 00072f44
>
> First col: pkts
> Second col: drops
> Third col: time_squeeze (by the polling loop)
> Last col: cpu_collsions (TX)
>
> eth0: UP Locked MII Full DuplexLink UP
> Admin up 20 hour(s) 23 min 44 sec
> Last input NOW
> Last output NOW
> 5min RX bit/s 91.5 M
> 5min TX bit/s 57.4 M
> 5min RX pkts/s 14261
> 5min TX pkts/s 12742
>
> Cheers.
> --ro
>
>
> takahashi@xxxxxxxxxxxx writes:
> >
> >
> > Since we are discussing this, heres more data ;->
> >
> >
> > ---------- Forwarded message ----------
> > Date: Wed, 3 Oct 2001 20:42:43 +0200
> > From: Robert Olsson <Robert.Olsson@xxxxxxxxxxx>
> >
> > Hello!
> >
> > Anyway. Uppsala University (35.000 students, 4000 emploees)
> > has now SMP/NAPI in production.
> > But reniceed ksoftirqd to -10. Kernel 2.4.10poll SMP.
> >
> >
> > USER PID %CPU %MEM SIZE RSS TTY STAT START TIME COMMAND
> > root 1 0.0 0.0 1224 512 ? S 18:08 0:03 init [5]
> > root 2 0.0 0.0 0 0 ? SW 18:08 0:00 (keventd)
> > root 3 74.2 0.0 0 0 ? RW< 18:08 100:02 (ksoftirqd_CPU0)
> > root 4 73.9 0.0 0 0 ? SW< 18:08 99:39 (ksoftirqd_CPU1)
> > root 5 0.0 0.0 0 0 ? SW 18:08 0:00 (kswapd)
> > root 6 0.0 0.0 0 0 ? SW 18:08 0:00 (bdflush)
> > root 7 0.0 0.0 0 0 ? SW 18:08 0:00 (kupdated)
> > root 52 0.0 0.0 796 192 ? S 18:08 0:00 voff
> > root 570 0.3 7.4 38696 38184 ? S 18:08 0:28 /sbin/gated
>
|