If code is split in sufficiently small crates compile times are not big of a deal for iterations. There is a faster development build and I would think that most time will be spent running the benchmark and checking perf to see processor usage dwarfing any time needed for compilation.