[Pgbigm-hackers] Remove USE_WIDE_UPPER_LOWER dependency

Back to archive index

Fujii Masao masao****@gmail*****
2013年 10月 10日 (木) 19:22:43 JST


On Thu, Oct 10, 2013 at 1:21 PM, Beena Emerson <memis****@gmail*****> wrote:
> Hello,
> I re-ran the tests a couple of times more. I found that there is no trend.
> I tried for show_bigm functions as well.
> Sometimes there is some increase(1-8%), sometimes no such increase is
> noticed. In few runs I have also seen a 0.5% decrease after patch applied.
>
> The procedure I followed was this.
> 1. Create query files to be executed
> query1.sql : SELECT * FROM bigm_tbl WHERE col LIKE
> likequery('postgresql.conf');
> query1.sql:  SELECT * FROM bigm_tbl WHERE col LIKE likequery('bigm');
> One more query for show_bigm was also used.
> SELECT show_bigm('max_standby_streaming_delay sets the maximum delay before
> canceling queries when a hot standby server is processing streamed WAL
> data.');
>
> 2. make install vanilla pg_bigm
>
> 3. Create a pgbench_db database, install pg_bigm., create a table and index
> and load data into it.
>
> 4. run pgbench
> pgbench -n -c 1 -T 300 -f query1.sql -d pgbench_db
> The same command was run 5 times each.
> This step was repeated for all query files.
>
> 5. remove the bigm and drop the database. Reboot system and install bigm
> with the patch applied. Repeat 3, 4.

What feature calls CPBIGM() most often? Text similarity search?
Normal text search? bigm_similarity()? Other?

It could be an idea to use that in the benchmark of the patch.

Regards,

-- 
Fujii Masao




Pgbigm-hackers メーリングリストの案内
Back to archive index