re: Mixed -current MP results

[Hauke Fath <hauke%Espresso.Rhein-Neckar.DE@localhost>]

At 6:59 Uhr +1000 10.5.2011, matthew green wrote:
>one comment wrt the sleep you're seeing, LOCKDEBUG is *extmerely* expensive
>with netbsd-5 and newer.  the slowdowns you're seeing are unfortunately to
>be expected.  in at least one case, we've observed at least a 90% drop of
>perf of an x86 quadcore system.

Ah, okay. Kind of to be expected with fine-grained locking spreading...
more locks, more debug code. My sparc isn't faster, yet, but I am less
worried.  ;)

How mandatory is LOCK_DEBUG these days?

        hauke


--
"It's never straight up and down"     (DEVO)