Call for performance evaluation: net.isr.direct (fwd)

Andrew Gallatin gallatin at cs.duke.edu
Thu Oct 13 16:55:03 PDT 2005


Garrett Wollman writes:
 > <<On Wed, 12 Oct 2005 17:17:12 -0400 (EDT), Andrew Gallatin <gallatin at cs.duke.edu> said:
 > 
 > > Right now, at least, it seems to work OK.  I haven't tried witness,
 > > but a non-debug kernel shows a big speedup from enabling it.  Do
 > > you think there is a chance that it could be made to work in FreeBSD?
 > 
 > I did this ten years ago for a previous job and was able to blow out
 > the stack very easily.

I haven't blown it out yet, but for that and other reasons, it seems
to be a bigger can of worms than it would be worth.

The interesting thing is that using the TSC timecounter rather than
ACPI-fast reduces the context switch latency enough so as to make the
TCP latency 25us when using a netisr thread.  25us is identical to
what I saw with the direct dispatch loopback hack.

Linux already takes care of syncing the TSC between SMP cpus, so we
know it is possible.  This seems like a much more doable optimization.
And it is likely to have other benefits..

Drew



More information about the freebsd-net mailing list