Messages in this thread | | | From | Arthur <> | Subject | Re: Cache thrashing? | Date | Sun, 19 Sep 1999 13:28:42 -0400 |
| |
On Sun, 19 Sep 1999 16:24:58 +0200 (CEST), Andrea Arcangeli <andrea@suse.de> wrote:
>I rewrote the reschedule_idle heuristics because the old 2.2.x heuristics >was buggy and made no sense to me. > >The 2.2.x version of my reschedule_idle rewrote is here (against 2.2.12 >but will apply cleanly against 2.2.13): > > ftp://ftp.suse.com/pub/people/andrea/kernel-patches/my-2.2.12/SMP-scheduler-2.2.11-E > >Please Arthur try it and feedback producing some performance number.
I'm running with it now. It may help some, but it's hard to tell.
The XMMS processes run less than 5.0 percent some of the time, but they still spike much higher, up to 15 percent or so. When they spike higher, they stay high for a while, then come back down after a minute or so.
Running UP, they never spike up like that.
My kernel is the mandrake helios version, which I think is the same as 2.2.13pre7. Here are all the Andrea patches I have applied to it:
>SA_INTERRUPT-shared-1 >SMP-boot-race-2.2.10-B >SMP-scheduler-2.2.11-E >bh-SMP-races-2.2.10-A >buffer-races-2.2.10-A >flush-tlb-i386-1 >no-swapout-2.2.10-B >ntp-SMP-races-1 >oom-2.2.12-I >wakeup_bdflush-2.2.10-A
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |