SLUB: Reduce default max order to zero There is no discernable performance benefit of order 1 pages over order 0. In fact netperf numbers are higher for order 0 which is likely due to the better cacheing of order zero pages in pcps and the avoidance of the zone lock. Tested on a UP system. Performance will likely suffer a bit under high contention in SMP systems or NUMA systems since this causes frequent uses of the list lock. But then the objects per slab are already pretty low since we cannot use higher order allocs like with antifrag in mm. Signed-off-by: Christoph Lameter --- mm/slub.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: slub/mm/slub.c =================================================================== --- slub.orig/mm/slub.c 2007-05-15 21:17:15.000000000 -0700 +++ slub/mm/slub.c 2007-05-15 21:22:01.000000000 -0700 @@ -158,7 +158,7 @@ static inline void ClearSlabDebug(struct /* * Small page size. Make sure that we do not fragment memory */ -#define DEFAULT_MAX_ORDER 1 +#define DEFAULT_MAX_ORDER 0 #define DEFAULT_MIN_OBJECTS 4 #else