Re: [sqlite] Big performance regressions since 3.4.0?

Or solve two problems by improving the algorithm for non-indexed GROUP BY queries: http://www.sqlite.org/cvstrac/tktview?tn=1809 D. Richard Hipp wrote: >Version 3.5.3 made a change to the way DISTINCT is processed. >Probably that change is making your particular case much slower. The >change

Re: [sqlite] Big performance regressions since 3.4.0?

improve your insert performance. --- On Mon, 1/5/09, D. Richard Hipp <d...@hwaci.com> wrote: > From: D. Richard Hipp <d...@hwaci.com> > Subject: Re: [sqlite] Big performance regressions since 3.4.0? > To: "General Discussion of SQLite Database" <sqlite-users@sql

Re: [sqlite] Big performance regressions since 3.4.0?

Version 3.5.3 made a change to the way DISTINCT is processed. Probably that change is making your particular case much slower. The change can be seen at: http://www.sqlite.org/cvstrac/chngview?cn=4538 This change was in response to grumbling on the mailing list

Re: [sqlite] Big performance regressions since 3.4.0?

Dan, Thanks, I sent a separate email containing the source and data for my test app since it's too big (5M) to post to the group. But I'm also posting just the code for the test app here, in case anyone wants to take a look. On my target hardware, the results I got for the most egregious

Re: [sqlite] Big performance regressions since 3.4.0?

On Dec 25, 2008, at 5:01 AM, Jason Freund wrote: > Hi, > > > > We recently upgraded from 3.4.0 to 3.6.5, and noticed that our little, > internal sqlite benchmark test program performed much worse on most > operations with the new version. We expected to see an improvement > because 3.6.1

[sqlite] Big performance regressions since 3.4.0?

Hi, We recently upgraded from 3.4.0 to 3.6.5, and noticed that our little, internal sqlite benchmark test program performed much worse on most operations with the new version. We expected to see an improvement because 3.6.1 advertises a performance enhancement with the lookaside buffer.