Re: Odd sorting behaviour

On Tue, Jul 20, 2004 at 10:06:08AM -0700, Josh Berkus wrote:
> Actually, quick question -- have you tried setting enable_mergjoin=false to
> see the plan the system comes up with? Is it in fact faster?

It is significantly faster -- 1200ms vs. 1900ms (on 7.4, at least). Some of
the merge joins are changed to nested loop joins, though, which probably
reduces the overall performance, so I guess there's more to gain if I can get
it to convert only that merge join to a hash join. The sum and multiplication
parts still take 400ms or so, though (is this normal? :-) ), so I guess
there's a lower limit :-)

I could of course post the updated query plan if anybody is interested; let
me know. (The data is still available if anybody needs it as well, of
course.)

/* Steinar */
--
Homepage: http://www.sesse.net/