Comment by alanctgardner3

In my experience, almost all overhead in Go programs will result from being bitten by memmove, alloc and GC. Profiling any program that uses interface{} and byte buffers will quickly expose you to how much the runtime loves to reallocate and copy objects. In one instance I got 10x throughput improvement by switching from a generic interface (take an interface{} and detect the type) to using typed methods and hand-rolling a shim that called the appropriate methods.