Note that after cranking up work mem you then need a query that can use it. Quickest way to use LOTS of memory is to do a lot of unconstrained joins.
select * from table a join table b on (true) join table c on (true) join table d on (true); Result set is size of a*b*c*d On Fri, Jun 19, 2015 at 7:21 AM, <holger.friedrich-fa-triva...@it.nrw.de> wrote: > Albe Laurenz wrote on Friday, June 19, 2015 12:56 PM: >> Holger Friedrich wrote: >> > So how do I craft a query that actually does use lots of memory? > >> You increase the parameter "work_mem". > >> You can do that globally in postgresql.conf or with SET for one session or >> with SET LOCAL for one transaction. > > Thank you to both Albe Laurenz and Andreas Kretschmer, who both gave the same > advice. It was spot-on: my query actually seems to be good for a > performance study, setting work_mem to various values causes the query > planner to either sort in memory or on disk, and of course this has some > impact on performance. > > (Sorry for erroneously asking for suggesting a "better" query, rather than > for tuning my configuration, which turned out to be the sensible thing to > do...) > > Best regards, > Holger Friedrich > > -- > Sent via pgsql-general mailing list (pgsql-general@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-general -- To understand recursion, one must first understand recursion. -- Sent via pgsql-general mailing list (pgsql-general@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
