| From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
|---|---|
| To: | Tomas Vondra <tv(at)fuzzy(dot)cz> |
| Cc: | "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: 9.5: Better memory accounting, towards memory-bounded HashAgg |
| Date: | 2014-08-22 19:25:12 |
| Message-ID: | CA+TgmoYGhOa7UA7ZfOmmfdeXe8PRiPykUUsPxCaO+tPR1W+JMQ@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Fri, Aug 22, 2014 at 2:13 PM, Tomas Vondra <tv(at)fuzzy(dot)cz> wrote:
> I don't think we really need to abandon the 'tracked' flag (or that we
> should). I think it was useful, and removing it might be one of the
> reasons why Robert now sees worse impact than before.
The version that introduced that flag had the same overhead as the
previous version that didn't, at least in my testing, so I'm not sure
it's at all useful. Part of the problem here is that the number of
instructions that you can afford to take to complete a memory
allocation is just really small, and so every one hurts. Memory
latency may be an issue as well, of course.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Alvaro Herrera | 2014-08-22 19:28:40 | Re: SKIP LOCKED DATA (work in progress) |
| Previous Message | Robert Haas | 2014-08-22 19:22:16 | Re: [GSoC2014] Patch ALTER TABLE ... SET LOGGED |