Another method for determining slow parts of code is to use benchmarks. Benchmarks can be used to test functions for average performance and can also run benchmarks in parallel. This can be useful when comparing functions or doing micro-optimizations for certain code, especially to see how a function implementation might perform when using it concurrently. For this recipe, we'll create two structures that both implement an atomic counter. The first will use the sync package, and the other will use sync/atomic. We'll then benchmark both the solutions.