I'm imagining that the index is only loading what it has to for each insert.
That results in head thrashing the disk when the file isn't cached. I'm going to profile this and see what pops out. I'm also going to use 3.7.9. Michael D. Black Senior Scientist Advanced Analytics Directorate Advanced GEOINT Solutions Operating Unit Northrop Grumman Information Systems ________________________________ From: [email protected] [[email protected]] on behalf of Fabian [[email protected]] Sent: Wednesday, November 09, 2011 3:30 PM To: General Discussion of SQLite Database Subject: EXT :Re: [sqlite] INDEX Types 2011/11/9 Black, Michael (IS) <[email protected]> OK...you're right...a reboot kills it. > I'm glad someone was able to reproduce this on Linux, ruling out the possibility it's a Windows-issue. > However, reboot again and add "select count(*) from a;" as the first line > of gendat2.sql > So if a simple SELECT COUNT(*) can speed up the same insert from 90 seconds to 7 seconds (including the count), does this confirm it's a sequential vs random reads problem? _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
