On Fri, 2014-09-19 at 21:14 -0400, Richard Hipp wrote:
> The latest SQLite 3.8.7 alpha version (available on the download page
> http://www.sqlite.org/download.html) is 50% faster than the 3.7.17 release
> from 16 months ago.  That is to say, it does 50% more work using the same
> number of CPU cycles.
> This performance gain is over and above the query planner improvements that
> have also been made.  We are constantly looking for new ways to run queries
> and adding those ways into the query planner.  For example, in the previous
> release, we added a new way to evaluate IN operators with non-constant
> right-hand-sides that was reported on this mailing list to make some
> queries run 5 times faster.
> The 50% faster number above is not about better query plans.  This is 50%
> faster at the low-level grunt work of moving bits on and off disk and
> search b-trees.  We have achieved this by incorporating hundreds of
> micro-optimizations.  Each micro-optimization might improve the performance
> by as little as 0.05%.  If we get one that improves performance by 0.25%,
> that is considered a huge win.  Each of these optimizations is unmeasurable
> on a real-world system (we have to use cachegrind to get repeatable
> run-times) but if you do enough of them, they add up.
> A full 10% of the performance gain has come since the previous release.
> There have been a lot of changes.  All our tests pass, and we still have
> 100% branch test coverage, so we are confident that we didn't break too
> much.  But your testing is an important part of our quality process.
> Please download a source archive or a DLL and give the latest alpha a
> whirl, and let us know if you encounter any problems.

That looks really promising; thanks for all this work.

Tristan, you have a comprehensive set of benchmarks for Evolution's
addressbook; is it possible for someone else to run those or would it
take more of your time to babysit than it would to run them yourself?

sqlite-users mailing list

Reply via email to