On Jan 15, 2008, at 9:34 AM, Stuart Henderson wrote:
I thought the performance improvement came from 4.1 with the
removal of per
packet interrupts.
http://www.openbsd.org/42.html
Huge performance improvements in the network stack, including:
# In pf, store routing table ID, queue ID etc directly in the packet
header mbuf instead of using mbuf tags (which use malloc'd memory).
This
yields a 100% improvement in pf performance.
# Packet forwarding can skip IPSEC stack if no IPSEC flows are
defined.
This yields a further 5% improvement in packet forwarding performance.
# Skip TCP/UDP/ICMP/ICMP6 checksumming when not necessary. This
yields a
further 10% improvement in pf performance.
Hmm. I'll do a test upgrade later this week, and once again try to
knock my latency up to something kind of insane.
The closest relevant information from plus42.html:
* Enable interrupt holdoff on DP83816 sis(4) chips. Significantly
improves
performance of such devices under load.
that doesn't help your 83815D.
I know this.