As mentioned in earlier mails, this will reduce the per-backend usage of memory by an amount which will be a fraction (single digit percentage) of (NBuffers * int) size . I have done pgbench/dbt2 runs and I do not see any negative impact because of this. Are there any other suggestions for measuring the backend memory footprint?
Simon Riggs wrote: > On Mon, 2006-11-27 at 14:42 -0500, Bruce Momjian wrote: > > Simon Riggs wrote: > > > int8 still seems like overkjll. When will the ref counts go above 2 on a > > > regular basis? Surely refcount=2 is just chance at the best of times. > > > > > > Refcount -> 2 bits per value, plus a simple overflow list? That would > > > allow 0,1,2 ref counts plus 3 means look in hashtable to find real > > > refcount. > > > > At two bits, would we run into contention for the byte by multiple > > backends? > > No contention, its a private per-backend data structure. That's why we > want to reduce the size of it so badly. > > -- > Simon Riggs > EnterpriseDB http://www.enterprisedb.com > > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend