On Mon, Apr 11, 2005 at 22:34:21 -0600,
Michael Fuhr <mike@fuhr.org> wrote:
> On Mon, Apr 11, 2005 at 11:40:01PM -0500, Bruno Wolff III wrote:
> > I was looking at:
> > http://candle.pha.pa.us/main/writings/pgsql/sgml/performance-tips.html
> >
> > And noticed in the first example that it is claimed the cost estimate
> > is 233 disk block reads. However, "233" doesn't appear in the explain
> > output shown. I think that "233" is supposed to match the "333" in
> > the explain output, but I am not 100% sure.
>
> This part?
>
> This is about as straightforward as it gets. If you do
>
> SELECT * FROM pg_class WHERE relname = 'tenk1';
>
> you will find out that tenk1 has 233 disk pages and 10000 rows.
> So the cost is estimated at 233 page reads, defined as costing
> 1.0 apiece, plus 10000 * cpu_tuple_cost which is currently 0.01
> (try SHOW cpu_tuple_cost).
>
> Doesn't that work out to
>
> (233 * 1.0) + (10000 * 0.01) = 233.0 + 100.0 = 333.0
>
> or am I missing something?
No, I was. I misread a comma as a period and when I scanned ahead to see
if I could find something similar in other examples I missed the part
where it added in the cpu cost. After I sent the message I looked some
more and spotted what I missed. I didn't have a copy back yet of the
original message, so my oops message wasn't threaded with the original.
