On Thu, 18 Aug 2005 23:30, Martin MOKREJŠ wrote: > But that is very short and does not affect the interpretation here.
Crap. The sync can take ages. > The throughput is clearly lower on 2.6 kernel and definitely the > CPU is in my eyes unnecessarily blocked... Why is the CPU in the > wait state instead of idle (this is teh problem on 2.6 series > but CPU is free on 2.4 series)? That's the main problem I think at the > moment. There is no wait state accounted for in 2.4 so you won't see it. Con > Con Kolivas wrote: > > On Thu, 18 Aug 2005 22:48, Martin MOKREJŠ wrote: > >>I think the problem here is outside afs. > >>Just doing this dd test but writing data directly to the ext2 > >>target gives same behaviour, i.e. on 2.4 kernel I see most of the > >>CPU idle but on 2.6 kernel all that CPU amount is shown as in > >>wait state. And the numbers from 2.4 kernel show higher throughput > >>compared to the 2.6 kernel (regardless the the PREEMPT or no PREEMPT > >>was used). > > > > Don't forget to include sync time. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/