On Fri, 18 Dec 1998, David S. Miller wrote:
> I can prove this. The performance counters on the Ultra, unless
> broken, support my claims completely. With idler IRQ forwarding on
> and the heuristics in the idler, system wide cache misses went down
> substantially (12% on average with about ~2 of error between test
> runs). [...]
on x86 we probably stay completely out of the main memory, not consuming
any 'global' bandwith even if tightly calling schedule(), at least if it's
a CPU with a decent MESI implementation. But even in this case there are
other global costs. (see my previous mail)
i certainly saw performance drop like a rock when i removed "hlt", but it
was noticable even when i added 'udelay(10)' to reduce the frequency of
re-schedules(). with current->need_resched we are certanily alot more
flexible now.
-- mingo
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
