Horton

Re: Pre-alloc ListCell's optimization

Stephen Frost <sfrost@snowman.net>

View thread
From: Stephen Frost <sfrost@snowman.net>
To: Robert Haas <robertmhaas@gmail.com>
Cc: pgsql-hackers@postgresql.org
Date: 2012-05-18T21:15:14Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Replace the parser's namespace tree (which formerly had the same

  2. Turns out that my recent elimination of the 'redundant' flatten_andors()

* Stephen Frost (sfrost@snowman.net) wrote:
> Alright, so I've done some pgbench's using all default configs with just
> a straight up './configure' and pgbench -S -T 300, 3 runs each and then
> averaged:
> 
> llist_opt: 9289 tps
> HEAD:      9286 tps

Ok, apparently part of the issue with the previous changes was that the
way copyList() worked it still ended up doing multiple palloc's, and
apparently that's used a lot.  Reworking that gave us a bit more
noticably improvement:

llist_opt: 9407 tps

Which gives us ~1.3% improvment.

The current patch still only pre-alloc's 8 ListCell's, but I've modified
it such that we might be able to use repalloc() to grow that number by 8
(or more..) instead of falling back to the regular palloc() for every
ListCell when we get more than 8 entries.

	Thanks,

		Stephen