Re: GiST index performance

Matthew Wakeling <matthew@flymine.org> writes:
> That sucks. However, as another observation, no calls to "gistfindnext"
> are recorded in the profile either, and that's in the same source file as
> "gistnext" which is recorded. Could it have been inlined?

Probably.

> Shouldn't inlining be switched off on a profiling build?

Why?  You generally want to profile the code's actual behavior, or as
near as you can get to observing that.  Defeating compiler optimizations
doesn't sound like something that -pg should do on its own.  If you
really want it, there's a switch for it.

> Oprofile scares me with the sheer number of options.

You can ignore practically all of them; the defaults are pretty sane.
The recipe I usually follow is:


Initial setup (only needed once per system boot):

sudo opcontrol --init
sudo opcontrol --setup --no-vmlinux

(If you need detail about what the kernel is doing, you need kernel
debug symbols and then specify them on the previous line)

Start/stop profiling

sudo opcontrol --start
sudo opcontrol --reset
... exercise your debug-enabled program here ...
sudo opcontrol --dump ; sudo opcontrol --shutdown

The test case should run at least a minute or two to get numbers with
reasonable accuracy.

Analysis:

opreport --long-filenames | more

opreport -l image:/path/to/postgres | more

if you really want detail:

opannotate --source /path/to/postgres >someplace

            regards, tom lane