A nicer nice scheduling
PATCH 1/1 A nicer nice scheduling
Hi,
please consider including this patch in the main kernel. It was proposed on 11/04/2001 by Rik van Riel ([test-PATCH] Re: [QUESTION] 2.4.x nice level)
This patch has been working great for me, I applied it to every new kernel out.
Without this patch, a nice=19 busy-looping process is given 15% of CPU cycles when there is a busy-looping nice=0 process.
Well, if one runs background CPU-consuming programs (dnetc), this patch makes a nice speed boost.
Please CC to me any replies.
include/linux/sched.h Sun Aug 12 12:19:16 2001 renamed from include/linux/sched.h Sun Aug 12 10:18:03 2001 +2 -1
@@ -305,7 +305,8 @@ * the goodness() loop in schedule(). */ long counter; - long nice; + short nice_calc; + short nice; unsigned long policy; struct mm_struct *mm; int has_cpu, processor;
kernel/sched.c Sun Aug 12 12:19:16 2001 renamed from kernel/sched.c Sun Aug 12 10:18:03 2001 +19 -1
@@ -680,8 +680,26 @@ struct task_struct *p; spin_unlock_irq(&runqueue_lock); read_lock(&tasklist_lock); - for_each_task(p) + for_each_task(p) { + if (p->nice <= 0) { + /* The normal case... */ p->counter = (p->counter >> 1) + NICE_TO_TICKS(p->nice); + } else { + /* + * Niced tasks get less CPU less often, leading to + * the following distribution of CPU time: + * + * Nice 0 5 10 15 19 + * %CPU 100 56 25 6 1 + */ + short prio = 20 - p->nice; + p->nice_calc += prio; + if (p->nice_calc >= 20) { + p->nice_calc -= 20; + p->counter = (p->counter >> 1) + NICE_TO_TICKS(p->nice); + } + } + } read_unlock(&tasklist_lock); spin_lock_irq(&runqueue_lock); }
Signature · 1 line
Roberto Ragusa robertoragusa at technologist.com
Quoted 7 lines · Roberto Ragusa
Hi,
please consider including this patch in the main kernel. It was proposed on 11/04/2001 by Rik van Riel ([test-PATCH] Re: [QUESTION] 2.4.x nice level)
Quoted 6 lines
This patch has been working great for me, I applied it to every new kernel out.
Without this patch, a nice=19 busy-looping process is given 15% of CPU cycles when there is a busy-looping nice=0 process. [...]
I think it's simpler to change NICE_TO_TICKS() macro in sched.c
Bye.