Postgres Pro
Downloads
Home   >   mailing lists

Re: A performance issue in ROW_NUMBER() OVER(ORDER BY NULL) [27 times slow than OVER()] V14.5 - Mailing list pgsql-general

From Kirk Wolak
Subject Re: A performance issue in ROW_NUMBER() OVER(ORDER BY NULL) [27 times slow than OVER()] V14.5
Date
Msg-id CACLU5mT2AU_Sk2Hx5xbCMET-b0_ph01CpVwXfpMXnruEVx_F_g@mail.gmail.com
Whole thread Raw
In response to Re: A performance issue in ROW_NUMBER() OVER(ORDER BY NULL) [27 times slow than OVER()] V14.5  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Tree view
On Sun, Feb 19, 2023 at 4:18 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
Kirk Wolak <wolakk@gmail.com> writes:
>   I have some converted code that uses this syntax.

Seems kinda dumb, but ...

>   The solution is to remove the ORDER BY NULL.  [since that is not
> sortable, should it be ignored?]
>   This does NOT SHOW UP with 1 million rows.

I don't see it at all.  Comparing your two test queries on released
branches, I see maybe 2x penalty for the ORDER BY NULL, not 30x.
(In HEAD there's only about 13% penalty.)  I wonder what PG version
you are testing.

                        regards, tom lane
Tom,
  I put V14.5 in the subject line (I could have made it more clear).
  It appears in new versions, as confirmed by Pavel, it is already addressed in some newer versions.

  Also, would it make sense to have EXPLAIN output the version of PG?  I think that might be useful,
because it becomes a COMMON next question?

Thanks,

Kirk Out!

 

pgsql-general by date:

Previous
From: Tom Lane
Date:
Subject: Re: A performance issue in ROW_NUMBER() OVER(ORDER BY NULL) [27 times slow than OVER()] V14.5
Next
From: Marc Millas
Date:
Subject: pg_reload_conf()