Thread

Re: spinlock contention

Robert Haas <robertmhaas@gmail.com> — 2011-06-29T01:17:32Z

On Tue, Jun 28, 2011 at 8:11 PM, Florian Pflug <fgp@phlo.org> wrote:
>> I wrote a little script to show to reorganize this data in a
>> possibly-easier-to-understand format - ordering each column from
>> lowest to highest, and showing each algorithm as a multiple of the
>> cheapest value for that column:
>
> If you're OK with that, I'd like to add that to the lockbench
> repo.

It's a piece of crap, but you're welcome to try to fix it up.  See attached.

With respect to the apparent lack of impact of the sleep loop, I can
only speculate that this test case doesn't really mimic real-world
conditions inside the backend very well.  There is a lot more going on
there - multiple locks, lots of memory access, etc.  But I don't
really understand it either.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company