Re: [PERFORM] 60 core performance with 9.3

2014-06-27 Thread Mark Kirkwood
On 27/06/14 21:19, Andres Freund wrote: On 2014-06-27 14:28:20 +1200, Mark Kirkwood wrote: My feeling is spinlock or similar, 'perf top' shows kernel find_busiest_group kernel _raw_spin_lock as the top time users. Those don't tell that much by themselves, could you do a hierarchical profile?

Re: [PERFORM] 60 core performance with 9.3

2014-06-27 Thread Andres Freund
On 2014-06-27 14:28:20 +1200, Mark Kirkwood wrote: > My feeling is spinlock or similar, 'perf top' shows > > kernel find_busiest_group > kernel _raw_spin_lock > > as the top time users. Those don't tell that much by themselves, could you do a hierarchical profile? I.e. perf record -ga? That'll a

[PERFORM] Can improve 'limit 1' ? with slow function

2014-06-27 Thread songtebo
The simplified scene: select slowfunction() from a order by b limit 1 is slow than select slowfunction() from ( select * from a order by b limit 1) if there are many records in table 'a' The real scene: function ST_Distance_Sphere is slow than ST_Distance, the query: SELECT ST_Distance_Sphere(