(This article will be split into multiple parts since it would be too long) Writing correct benchmarks is hard, and it's a craft. Benchmarks try to test a piece of code running in isolation which is very hard, and the results will be mainly in milliseconds or even in smaller units of time. Consider how hard it is to measure such timings correctly and filter out the noise and other system components running at the time that might contribute to the benchmark result. Performance Testing APIs or Large s...