On Tue, Aug 02, 2005 at 04:09:17PM +1000, Nick Piggin wrote:
> I have a patch here which I still need to do more testing with,
> which might help performance on HT systems.
>
> I found that idle siblings could cause SMP and NUMA balancing to
> be too aggressive in some cases.
> --
> If an idle sibling of an HT queue encounters a busy sibling, then
> make higher level load balancing of the non-idle variety.
Makes sense and patch looks good. Please push this minor comment fix along
with your patch. Thanks.
--- linux-2.6.13-rc4/kernel/sched.c~ 2005-08-02 13:36:34.804764128 -0700
+++ linux-2.6.13-rc4/kernel/sched.c 2005-08-02 13:38:00.689707632 -0700
@@ -2316,7 +2316,9 @@
if (j - sd->last_balance >= interval) {
if (load_balance(this_cpu, this_rq, sd, idle)) {
- /* We've pulled tasks over so no longer idle */
+ /* We've pulled tasks over so no longer idle
+ * or one of our SMT sibling is not idle.
+ */
idle = NOT_IDLE;
}
sd->last_balance += interval;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
