Re: [ckrm-tech] Re: [RFC][PATCH 3/9] CPU controller - Addstimeslice scaling

[Mike Galbraith]

On Fri, 2006-04-21 at 20:50 +0900, Naoaki MAEDA wrote:

> It was not good explanation. Let me restate that.
> The effect of shortening timeslice is to let the task be expired soon
> by shortening
> its remainder timeclice, so it still works even if the task consome very small
> timeslice at one time. However, expired TASK_INTERACTIVE tasks will be requeued
> to the active for a while by the scheduler, so shortening timeslice
> doesn't work well for
> TASK_INTERACTIVE tasks.

Yeah, understood.  This shortening of timeslice I think is generally
bad, (hmm...) though in an environment where preemption is rampant, this
shortening of slice should lead to a throughput gain for low ranking
_interdependent_ tasks.  Is that what you're trying to accomplish?  To
reduce latency at the bottom in the face of preemption from above?

	-Mike

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/