Horton

Re: Binary search in fmgr_isbuiltin() is a bottleneck.

tushar <tushar.ahuja@enterprisedb.com>

View thread View on postgr.es
From: tushar <tushar.ahuja@enterprisedb.com>
To: Andres Freund <andres@anarazel.de>, pgsql-hackers@postgresql.org
Date: 2017-09-28T06:05:11Z
Lists: pgsql-hackers 
On 09/27/2017 05:29 PM, tushar wrote:
> After discussion with Jeevan Ladhe, we created a sql query which 
> contain lots of inbuild function and tested that against pgbench    
> with master v/s patch and found an improvement

I tested it again and found around +2% improvement

./pgbench -c 8 -j 8 -f /tmp/mytest.sql -T =TIME


After taking Median of 3 run  -

Case 1 – TIME=300

PG HEAD =>tps = 7831.999245 (excluding connections establishing)
PG HEAD+patch =>tps= 8008.895177 (2.26+% vs. head)

Case 2- TIME=500

PG HEAD =>tps = 7817.781756 (excluding connections establishing)
PG HEAD+patch =>tps= 8050.410040(2.98+% vs. head)

Case 3- TIME=1000

PG HEAD =>tps = 7817.173640 (excluding connections establishing)
PG HEAD+patch => tps= 8011.784839(2.48+% vs. head)

Case 4-TIME=1500

PG HEAD =>tps = 7764.607133 (excluding connections establishing)
PG HEAD+patch =>tps= 8013.421628(3.20+% vs. head)
-- 

regards,tushar
EnterpriseDB  https://www.enterprisedb.com/
The Enterprise PostgreSQL Company

Commits

  1. Replace binary search in fmgr_isbuiltin with a lookup array.

  2. Add inline murmurhash32(uint32) function.