On Fri, Dec 17, 2010 at 7:29 PM, Richard Hipp <d...@sqlite.org> wrote:
> If you have any large queries that you can run on windows using the > command-line shell, I would appreciate you timing those queries using the > new shells from the download page, both with "-heap 300M" and without it, > and letting me know about any performance differences you see. > > Richard, I tried several queries and the ones that was FTS3-related showed about 40-50% improvement. Other artificial tests with large intermediate storage (forced to be in memory by PRAGMA temp_store=2) didn't appear better. Personally it would be great if sqlite has its own heap manager just to be accurate also in this area and to rely on its own polished code. But I thought recently about cases when heap manager can affect the performance in general. - If a large query reads random data not present in sqlite/system cache, It contributes not much - If additional storage needed, this depends on the storage chosen (disk/memory). In case of disk, it's also not so important. - full-test is a good example where it matters. I think this is because in order to return smaller result it sometimes needs more memory (for example when resolving 'and' or phrase search) Just curious. I'm not so familiar with sqlite internals. Can someone mention some other cases when sqlite really needs to allocate memory frequently in order to affect the overall performance? Max Vlasov _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
