On Sun, 24 Dec 2006, Robert Watson wrote:
From the perspective of optimizing these particular paths, small packet
sizes
best reveal processing overhead up to about the TCP/socket buffer layer on
modern hardware (DMA, etc). The uni/bidirectional axis is interesting
because it helps reveal the impact of the direct dispatch vs. netisr dispatch
choice for the IP layer with respect to exercising parallelism. I didn't
explicitly measure CPU, but as the configurations max out the CPUs in my test
bed, typically any significant CPU reduction is measurable in an improvement
in throughput. For example, I was easily able to measure the CPU reduction
in switching from using the socket reference to the file descriptor reference
in sosend() on small packet transmit, which was a relatively minor functional
change in locking and reference counting.
Be careful with micro-optimizations. I saw a single change (adding
about 1K in unrelated code that is never executed) give a pessimization
of 15% for tx bge (from 360 kpps to 300 kpps). Before that I was
trying harder than now to find optimizations involving avoiding copying,
and thought that I had increased the speed from 330 kpps to 360 kpps
by removing things, but I may have just increased the speed by moving
cache phenomena. The phenomena in this case seem to be related to
instructions more than data and I suspect that they are very MD. The
machine that has them doesn't support APIC or ACPI, so hwpmc cannot
do anything useful on it.
Bruce
_______________________________________________
cvs-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/cvs-all
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
