Ian Rogers wrote: > The Dacapo benchmarks home page [1] does comparative performance > regressions with 3 months of data. They have DRLVM SVN in their results.
Thanks for pointing that out Ian. I've grabbed a couple of charts to illustrate my talk. > Jikes RVM has a framework for continuous regression and performance > testing called cattrack [2] and we provide numerous test harnesses for > running tests and then uploading the results into cattrack. Its all open > source [3] so maybe this solution could be shared by Jikes RVM and Harmony. Cool. We are already running a number of test scenarios and suites in the Harmony harness. I see no problem with having more and different testing. If you want to post pass/fail results based on Harmony code we have an [EMAIL PROTECTED] list for that purpose. Regards, Tim > [1] http://www.dacapobench.org/ > [2] http://jikesrvm.anu.edu.au/cattrack/ > [3] http://jikesrvm.svn.sourceforge.net/viewvc/jikesrvm/cattrack/trunk/ > > Aleksey Shipilev wrote: >> Yeah, it would be interesting to see whether we doing systematic >> performance testing at all? >> I have very old data (like May 2008) on >> Dacapo/SPECjbb2005/SPECjvm2008/Eclipse. >> >> On Thu, Oct 30, 2008 at 3:39 PM, Tim Ellison <[EMAIL PROTECTED]> wrote: >> >>> Do we have any up to date performance numbers published? >>> >>> There have been posts on this list about people running SPEC benchmarks, >>> and recent work on the JIT. >>> >>> The website [1] shows some very old results. >>> >>> [1] http://harmony.apache.org/performance.html >>> >>> Regards, >>> Tim >>> >>> > >
