Packet passing performance study on exotic hardware.
dgilbert at dclg.ca
Fri Oct 8 09:18:17 PDT 2004
>>>>> "Mike" == Mike Tancsa <mike at sentex.net> writes:
Mike> At 10:08 AM 08/10/2004, David Gilbert wrote:
>> Right out of the box, FreeBSD 5.3 (with polling) passed about 200
>> kpps. net.isr.enable=1 increased that without polling to about 220
Mike> Did you have kern.polling.idle_poll at 0 or 1 ? In my tests a
Mike> few weeks ago this seemed to make a difference, but the load avg
Mike> gets messed up. Also, HZ does seem to make a difference at
Mike> least in my tests on BETA5.
I can confirm the HZ not making a sizable difference (although I
believe it cuts polling latency under ligher load, so we used 10000 by
default and we tested 1000).
Idle_poll is default 1, I'm not positive we tested 0. I don't think
there is much idle time here.
|David Gilbert, Independent Contractor. | Two things can only be |
|Mail: dave at daveg.ca | equal if and only if they |
|http://daveg.ca | are precisely opposite. |
More information about the freebsd-current