RE: [Lse-tech] Re: [patch] scheduler fix for 1cpu/node case

From: Luck, Tony (tony.luck@intel.com)
Date: Wed Jul 30 2003 - 11:34:51 EST


> As for idle balances, we may be able to go a step further:
> follow the range rules, but do a more aggressive/frequent search.

Be cautious about how much work you do here ... while you can burn
as much cpu time as you like on an idle cpu without affecting anything,
you need to be sure that your activities don't burn too much bus
bandwidth, or cause cache lines to ping-pong around the machine. The
classic case of this has been seen while one cpu is trying to boot,
and the other 31 idle cpus beat the bus to death looking to see
whether they can "help".

-Tony
-
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/



This archive was generated by hypermail 2b29 : Thu Jul 31 2003 - 22:00:46 EST