This leads to slower results in ParseBenchmark (~ 5%) but with parsing taking only 10% of the total time in the main benchmark, the potential for performance regressions is severely limited. We see an improvement of overall benchmark times, both single-threaded and multi-threaded with a shared parse cache, in the main benchmark.
This leads to slower results in ParseBenchmark (~ 5%) but with parsing taking only 10% of the total time in the main benchmark, the potential for performance regressions is severely limited. We see an improvement of overall benchmark times, both single-threaded and multi-threaded with a shared parse cache, in the main benchmark.