On Aug 3, 2009, at 2:06 PM, Kosenko Max wrote: > > > Dan Kennedy-4 wrote: >> McObject CEO Steve Graves points out that because of limits of the >> API >> they were using, SQLite performs each INSERT and DELETE in the test >> in a >> separate transaction. So the reported times for these tests may be >> more of >> a measure of the speed of the media than SQLite itself. >> >> The body of the report shows that SQLite scans (iterates through) >> rows >> more quickly than Perst does (about 1.5 times faster). But Perst >> can look >> up individual objects more quickly (about 6 times faster). The >> comments >> in the report indicate that SQLite was not using pre-compiled SQL or >> read-only transactions for the object-lookup test, so that gap could >> probably be closed some by better use of the SQLite API. Perhaps more >> importantly though, Perst is an Object DB, so it is caching the >> actual >> objects in-memory, whereas SQLite is only caching database pages. I >> haven't much experience with them, but I think Object DBs are usually >> faster for this kind of task. YMMV. > > While this is offtopic, I just want to comment that there could be a > problems with any tests around, so you can propose changes to test > code if > you feel it could be done better.
I have no issue with the content of the report or formulation of the tests. I think the folks at McObject have done a good job of testing the two systems and shown real integrity by explaining the results. Most commercial operations are much less forthcoming. > Read-only transaction and skipping single preparation wouldn't speed- > up > SQLite in that test. Earlier I just quoted the conclusions of the McObject report. Maybe I misunderstood. But now that I have read the benchmark code, I'm curious. Why is the SQL not being recompiled for each query? Is there some kind of compiled query cache hiding behind the [db.query()] call in Android? Why would it not help to use a read-only transaction? Is SQLite using exclusive-locking mode? In the tests I've been working on today I've found that read-only transactions speed up SQLite considerably for this kind of workload, even when using exclusive-locking mode. Are you by any chance the author of the report I'm reading? Dan. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
