Re: Spinlocks, yet again: a new test case - Mailing list pgsql-hackers

From Gavin Sherry
Subject Re: Spinlocks, yet again: a new test case
Date
Msg-id Pine.LNX.4.58.0509120937080.14619@linuxworld.com.au
Whole thread Raw
In response to Spinlocks, yet again: a new test case  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Sun, 11 Sep 2005, Tom Lane wrote:

> The files for the updated test case are attached if anyone else wants
> to try it.  They are:
>
>     test_setup.sql    Run this to create the test tables
>
>     test_run.sql    A version of the test query that will run
>             pretty nearly forever (until you kill it)
>
>     test_run_small.sql    A one-minute-or-so test query
>
>     startn.sh    A shell script to run N copies of
>             test_run_small.sql in parallel.
>
> You need this one-time setup:
>
>     createdb test
>     psql -f test_setup.sql test
>
> To perform testing, do "time ./startn.sh N" for N=1,2,... up to however


With HEAD on a 4-way Xeon, I'm seeing:

1: 56  2: 120 4: 230

Removing the the non-locking test and branch, which you suggested
down-thread:

1: 55: 2: 111: 4: 207

--

Gavin


pgsql-hackers by date:

Previous
From: Kurt Roeckx
Date:
Subject: Re: Spinlocks, yet again: analysis and proposed patches
Next
From: Oliver Jowett
Date:
Subject: Re: statement logging / extended query protocol issues