> A recent thread in comp.arch discusses context switch times for various
> operating systems on modern hardware.  Someone who works at QNX just posted a
> list of measurements he's taken; it shows context switch times of 210ms and
> 380ms for 386BSD 0.1 and NetBSD on the same hardware.

(1) 380/210 != 2.  (it's 1.809)
(2) "on the same hardware" is probably not the case.  Note that the
	results were "collected from various responses on the net,"
	i.e. the CPU type may be the same, but the motherboard, et al.
	may be completely unrelated.  This can make a significant
	difference.
(3) our switch does a fair amount more, in terms of keeping more
	careful process accounting information, than the 386BSD
	switch() did.  In particular, mi_switch() calls microtime()
	twice, and that's not a cheap function.

I don't actively measure our kernel, though i probably should.
If you want to, go for it.  8-)

As it is, though, i wouldn't take that message as any cause for
alarm.  I'd want to run NetBSD on _exactly_ the same hardware
as the other "same hardware" systems before i would believe it.



cgd
who shouldn't be writing e-mail this late at night...  8-)

------------------------------------------------------------------------------