> main issue is interrupt count does not go below 4000. . .need to dig more > into this, probably to compose a mail to the list :), but it does not hurt > too much, at least for now.
that shouldn't be happening unless you're doing a lot of disk/network io. if you're using 9atom, one would expect ~1100 interrupts/sec since HZ is set to 1000. this makes timing work better and doesn't cost much performance. in 9atom, /dev/irqalloc has an extra field that is the count of total interrupts. i'm guessing that you have a ringing ide interrupt. but irqalloc should let you know who's the culprit. - erik
