Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Magic 200MHz (more on halt)

Author: Robert Hyatt

Date: 13:52:34 05/27/03

Go up one level in this thread




I'm looking at my linux kernel right now.  The SMP kernel defaults to a
"poll_idle()" function on this box.  It _never_ executes a halt instruction.

I'm looking into why this is, but whatever the reason, it is the case on
this particular box.

More after I finish running thru the scheduler code.

The comments suggest this is done because
/*
 * On SMP it's slightly faster (but much more power-consuming!)
 * to poll the ->need_resched flag instead of waiting for the
 * cross-CPU IPI to arrive. Use this option with caution.
 * Note: on pIV systems this actually saves power!
 */

That last is a strange comment that suggests that a halt burns more power
than a poll loop.  That doesn't compute in my book, but without a direct
citation, I certainly can't contradict it.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.