Re: [patch] [sched] load-balancer improvements, 2.5.3-pre5

[Richard Guenther]

>  static void smp_tune_scheduling (void)
>  {
> @@ -957,9 +959,13 @@
>  		cacheflush_time = (cpu_khz>>10) * (cachesize<<10) / bandwidth;
>  	}
>
> +	cache_decay_ticks = (long)cacheflush_time/cpu_khz * HZ / 1000;
> +
>  	printk("per-CPU timeslice cutoff: %ld.%02ld usecs.\n",
>  		(long)cacheflush_time/(cpu_khz/1000),
>  		((long)cacheflush_time*100/(cpu_khz/1000)) % 100);
> +	printk("task migration cache decay timeout: %ld msecs.\n",
> +		(cache_decay_ticks + 1) * 1000 / HZ);
>  }

Isnt it better for such randomly(?) choosen numbers like 1000 and 100
which you use to divide / modulo to choose them as a near power of two?
Like 1024 for / 1000 and 128 for the */% 100 above? For correctness just
change cpu_khz to be 1024*hz, not 1000*hz.

Richard.

--
Richard Guenther <[e-mail]>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/