> You asked me to re-do the tests I had run to check the GEQO
> threshold. With today's sources I get these runtimes:
>
> # Tables: 7 8 9 10 11
>
> std, with indexes 1.89 2.26 5.41 24.21 90.04
> GEQO, with indexes 3.73 6.98 16.98 43.81* 21.45
> std, no indexes 1.60 1.79 4.78 15.75 84.17
> GEQO, no indexes 3.30 6.04 15.52 18.73 22.35
>
> So, (a) the number of indexes is still not very relevant,
> and (b) it looks like we oughta kick over to GEQO at 11 tables.
>
> WHat were we at before? 6 or 8 I think? Nice job :-)
>
> regards, tom lane
>
> * That number is obviously out of line ... probably it was
> skewed by some other activity on my system at the time.
> I'm too lazy to repeat the measurement however.
>
Done. Thanks for doing the checking.
Doesn't seem like much of an increase, but it is N!, so I guess it is:
> 6! 720> 8! 40320> 11! 39916800
-- Bruce Momjian | http://www.op.net/~candle maillist@candle.pha.pa.us | (610)
853-3000+ If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill,
Pennsylvania19026