Re: [sqlite] Simple Group By slowing queries by 6x

2013-06-17 Thread Keith Medcalf
> boun...@sqlite.org] On Behalf Of Iván de Prado > Sent: Monday, 17 June, 2013 05:30 > To: sqlite-users@sqlite.org > Subject: [sqlite] Simple Group By slowing queries by 6x > > I have query where adding a simple "group by" by date is slowing the query > too much, from my po

Re: [sqlite] Simple Group By slowing queries by 6x

2013-06-17 Thread Clemens Ladisch
Iván de Prado wrote: > SELECT ddate, sum(hits) from a_idadvertiser_site_hostname_bench where ddate < > '2013-08-01' group by ddate; > > SCAN TABLE a_idadvertiser_site_hostname_bench (~33 rows) > USE TEMP B-TREE FOR GROUP BY > > [...] means that this query is running almost 6 times slower than

Re: [sqlite] Simple Group By slowing queries by 6x

2013-06-17 Thread Iván de Prado
Seems that using pragma temp_store=2 improves the speed. It now tooks 20 seconds. So now the penalty time is 9 seconds. It would possible to improve it?. Regards. Iván 2013/6/17 Iván de Prado > I have query where adding a simple "group by" by date is slowing the query > too

[sqlite] Simple Group By slowing queries by 6x

2013-06-17 Thread Iván de Prado
I have query where adding a simple "group by" by date is slowing the query too much, from my point of view. The following query: SELECT ddate, sum(hits) from a_idadvertiser_site_hostname_bench where ddate < '2013-08-01'; runs in 12 seconds. Approximately 423,327 rows scanned per second. 1 row