On Sat, 2008-05-31 at 11:53 -0700, Josh Berkus wrote:
> Simon,
>
> > There is an optimum for each specific sort.
>
> Well, if the optimum is something other than "as much as we can get", then we
> still have a pretty serious issue with work_mem, no?
Depends upon your view of serious I suppose. I would say it is an
acceptable situation, but needs further optimization. I threw some ideas
around on Hackers around Dec/New Year, but I don't have time to work on
this further myself in this dev cycle. Further contributions welcome.
--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support
