On Tue, 26 Apr 2005, Tom Lane wrote:
> Pavel Stehule <stehule(at)kix(dot)fsv(dot)cvut(dot)cz> writes:
> > I tested bitmap scan and maybe I didnt find good examples, but with bitmap
> > scan is slower than hashjoin. Only when I use non otiptimized SELECT bps
> > was little bit faster. All my SELECTs are equal.
> Bitmap scans can't possibly be any faster for cases where the indexscan
> only fetches one row, which is true of all your test cases AFAICS.
yes, it's true. I found some selects where the benefit of bitmap scans is
more clearly. There is only one small problem - optimizer didn't
have to choose plan with bitmap scan in my examples.
Thank you for explication,
In response to
pgsql-hackers by date
|Next:||From: Simon Riggs||Date: 2005-04-27 07:45:10|
|Subject: Re: [HACKERS] Bad n_distinct estimation; hacks suggested?|
|Previous:||From: Greg Stark||Date: 2005-04-27 05:59:30|
|Subject: Re: [PERFORM] Bad n_distinct estimation; hacks suggested?|