I realized the navigation benchmarks should measure CPU time spent in Performance.TIME_NAVIGATION_PROCESS. This requires modifying the benchmarks infrastructure to support a new measurement type (we have one for Performance.TIME_PROCESS already). Otherwise, we're just measuring setup time for each benchmark (which is important too, but should not be the only measurement of navigation performance).
Maybe we can map it to PROCESS in those benchmarks somehow, so that we don't need yet another column on the benchmarks website. I doubt any benchmark will need to measure both Performance.TIME_PROCESS and Performance.TIME_NAVIGATION_PROCESS at the same time.
I realized the navigation benchmarks should measure CPU time spent in
Performance.TIME_NAVIGATION_PROCESS. This requires modifying the benchmarks infrastructure to support a new measurement type (we have one forPerformance.TIME_PROCESSalready). Otherwise, we're just measuring setup time for each benchmark (which is important too, but should not be the only measurement of navigation performance).Maybe we can map it to PROCESS in those benchmarks somehow, so that we don't need yet another column on the benchmarks website. I doubt any benchmark will need to measure both
Performance.TIME_PROCESSandPerformance.TIME_NAVIGATION_PROCESSat the same time.