Benchmark results might vary based on a number of factors. I tend to select a fixed random seed so the test run is repeatable. Also, I disable Intel's Turbo Boost feature, so it doesn't vary the CPU frequency with overclocking. I run the benchmark loop multiple times, so the JIT is warmed up. With these tweaks, you get more consistent and meaningful results.
Benchmark results might vary based on a number of factors. I tend to select a fixed random seed so the test run is repeatable. Also, I disable Intel's Turbo Boost feature, so it doesn't vary the CPU frequency with overclocking. I run the benchmark loop multiple times, so the JIT is warmed up. With these tweaks, you get more consistent and meaningful results.