On 9/6/07, Sam Fourman Jr.  wrote:

quoted text
> hello misc@

> from the page http://www.openbsd.org/42.html , one of the changes made

> to OpenBSD 4.2 is

>

> A change in the way the kernel random pool is stirred greatly

> increases performance with network interface cards that support

> interrupt mitigation, especially on architectures where reading the

> clock is expensive (such as amd64).

>

> What would be some Examples of Network Cards that Support "interrupt mitigation"

>

> I guess on this Subject I need educated because I am not all together

> sure what interrupt mitigation is and why I want it.

I'm not the right person to try to explain it, but I have generally

been interested in seeing where it crops up and what it's associated

with.
The various /plusXY.html pages often have notes indicating drivers for

which interrupt mitigation has been enabled/enhanced, and you can see

what generally has it. For example, from plus38.html:
"Add microcode to support interrupt mitigation on fxp(4) 82551 F

stepping chipset. Big performance boost."
plus39.html:

"Fix for the nfe(4) interrupt mitigation code."
...or a CVS commit about enabling it in a National Semiconductor chip:

http://marc.info/?l=openbsd-cvs&m=118037514315592&w=2
...after which Soekris boxen with the DP83816 chips could experience

better throughput at lower interrupt rates, good since the poor

thing's CPU is a limiting factor. It's fun to look for this kind of

stuff in the CVS commit logs, e.g.
http://marc.info/?l=openbsd-cvs&w=2&r=1&s=%22interrupt+mitigation%22&q=b

http://marc.info/?l=openbsd-cvs&w=2&r=1&s=%22interrupt+holdoff%22&q=b

http://marc.info/?l=openbsd-cvs&w=2&r=1&s=interrupt+coalesce&q=b

and other places:

http://www.openbsd.org/papers/cuug2007/mgp00016.txt
DS