On 05/04/2011 04:26 AM, Jonathan "Duke" Leto wrote: > Is there any way you can generate similar data for small, specific subsets of > the spectests? Such as, subsets for OO, String, Float and Integer performance. > That way, we can pinpoint which subsystems are having performance regressions.
While I could probably run the tests, I don't have any idea how to chose such spectests, because I don't know which operations dominate the execution times of which spectests. Sure, we have spectests that do lots of numeric operations, but I have no idea if the execution time of the numeric operations isn't dwarfed by that of dispatching the numerical operators, lookup in lexicals and so on. If you want to test particular subsystems, performance tests in PIR (or maybe NQP/NQP-rx) look like the better alternative to me, at least until we have better profiling tools. Cheers, Moritz _______________________________________________ http://lists.parrot.org/mailman/listinfo/parrot-dev
