Messages in this thread | | | Date | Sun, 27 Jul 2003 12:26:56 +0200 (CEST) | From | Ingo Molnar <> | Subject | Re: [patch] sched-2.6.0-test1-G3, interactivity changes, audio latency |
| |
On Sat, 26 Jul 2003, Guillaume Chazarain wrote:
> BTW, with your nanosecond resolution, mplayer is also max interactive.
good.
> Don't you find it strange to need a nanosecond resolution to evaluate a > simple integer in a [-5, 5] range?
no, i dont find it strange, as the -5..5 integer range is only the end result. What we really want is an accurate sleep average, the integral of sleep times done over the last N seconds. Given that tasks can schedule at very high frequencies these days, this needs accurate measurements. I used to hope that a 1 msec sampling frequency would to be enough for this, but when we fix the statistics cornercase you mention, another cornercase pops up. And it's not _that_ hard to use the cycle counter and still have a reasonably fast algorithm, anyway.
Ingo
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |