On Mon, 2007-07-23 at 12:35 -0400, Tom Lane wrote:
> "Simon Riggs" <simon(at)2ndquadrant(dot)com> writes:
> > Well, I discover there is an allocation of 8232 (inflation...) made once
> > per statement by a memory context called... ExecutorState. Still not
> > sure exactly which allocation this is, but its definitely once per
> > statement on pgbench, which should narrow it down. Plan, query etc?
> Are you working with stock sources? The only allocation exceeding 1K
> that I can see during pgbench is BTScanOpaqueData, which is 6600 bytes.
> (Checked by setting a conditional breakpoint on AllocSetAlloc.) The
> path that allocates a single-chunk block is never taken at all.
I do have the bitmap patch currently applied, but it doesn't touch that
part of the code.
(gdb) p size
$1 = 8232
(gdb) p sizeof(int)
$2 = 4
(gdb) p sizeof(BTScanPosData)
$3 = 4104
Since my notes say I got 8228 last year, seems reasonable.
In response to
pgsql-performance by date
|Next:||From: Tom Lane||Date: 2007-07-23 18:19:20|
|Subject: Re: User concurrency thresholding: where do I look? |
|Previous:||From: Paul van den Bogaard||Date: 2007-07-23 17:42:06|
|Subject: Re: disable archiving|