Horton

Re: Default setting for enable_hashagg_disk

Jeff Davis <pgsql@j-davis.com>

View thread View on postgr.es
From: Jeff Davis <pgsql@j-davis.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: Tomas Vondra <tomas.vondra@2ndquadrant.com>, Tom Lane <tgl@sss.pgh.pa.us>, Alvaro Herrera <alvherre@2ndquadrant.com>, David Rowley <dgrowleyml@gmail.com>, Stephen Frost <sfrost@snowman.net>, Andres Freund <andres@anarazel.de>, Bruce Momjian <bruce@momjian.us>, Robert Haas <robertmhaas@gmail.com>, Justin Pryzby <pryzby@telsasoft.com>, Melanie Plageman <melanieplageman@gmail.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-07-29T17:13:23Z
Lists: pgsql-hackers, pgsql-docs 
On Sat, 2020-07-25 at 17:52 -0700, Peter Geoghegan wrote:
> BTW, your HLL patch ameliorates the problem with my extreme "sorted
> vs
> random input" test case from this morning [1] (the thing that I just
> discussed with Tomas). Without the HLL patch the sorted case had 2424
> batches. With the HLL patch it has 20. That at least seems like a
> notable improvement.

Committed.

Though I did notice some overhead for spilled-but-still-in-memory cases
due to addHyperLogLog() itself. It seems that it can be mostly
eliminated with [1], though I'll wait to see if there's an objection
because that would affect other users of HLL.

Regards,
	Jeff Davis

[1] 
https://www.postgresql.org/message-id/17068336d300fab76dd6131cbe1996df450dde38.camel@j-davis.com

Commits

  1. Add hash_mem_multiplier GUC.

  2. HashAgg: use better cardinality estimate for recursive spilling.

  3. Remove hashagg_avoid_disk_plan GUC.

  4. Doc fixup for hashagg_avoid_disk_plan GUC.

  5. Rework HashAgg GUCs.

  6. Disk-based Hash Aggregation.

  7. Implement partition-wise grouping/aggregation.

  8. Defer creation of partially-grouped relation until it's needed.