Jan Urbański <wulczer@wulczer.org> writes:
> I'm pleasantly surprised that the SA code as it stands today, setting
> the equlibrium factor to 8 and temperature reduction factor to 0.4, the
> query takes 1799.662 ms in total.
Cool.
> With the default values it runs
> forever, but I long discovered that defaults taken from the original
> paper are not well suited for my PG implementation (I could plug my MSc
> thesis here, but I'm way too shy for that). 8/0.4 are values where I got
> better results than GEQO for Andres' monster-query.
Hmmm ... "runs forever" is a bit scary. One of the few good things I
can say about GEQO is that it will terminate in a reasonable amount of
time for even quite large problems. I would like to think that SA will
also have that property. I thought that the annealing approach was sure
to terminate in a fixed number of steps? Or did you mean that the
planner terminated, but produced a horrid plan?
regards, tom lane