In run-perf-tests, we use run BYEBench 4 times with different instances of DRT/WTR to get a stable number and account for differences in things like ALSR and RNG initialization. When the benchmark is ran inside a browser, we can't account for these effects but we could at least run for 20 iterations and get a more stable number.
Created attachment 231248 [details] Patch
Created attachment 231249 [details] Dumped up the version number to 0.11
Comment on attachment 231249 [details] Dumped up the version number to 0.11 Clearing flags on attachment: 231249 Committed r168732: <http://trac.webkit.org/changeset/168732>
All reviewed patches have been landed. Closing bug.