Messages in this thread | | | Subject | Re: Scheduler regression: Too frequent timer interrupts(?) | From | Andi Kleen <> | Date | Sat, 18 Apr 2009 09:59:11 +0200 |
| |
"Chris Friesen" <cfriesen@nortel.com> writes: > > Given that we're talking about tens of usecs of duration, statistical > sampling may not work all that well.
He could just use oprofile with a small period. The minimum period for CPU_CLK_UNHALTED is 6000 (and it could be probably lowered by tweaking the events) oprofile counters on x86 can be configured to only count in ring 0 So you would get an entry every 6000 ring 0 cycles. That should give a pretty good picture where the kernel is spending time.
-Andi
-- ak@linux.intel.com -- Speaking for myself only.
| |