Today the time complexity of algorithm submissions is measured based on milliseconds of execution. This is both non-deterministic and less than totally scientific. I can get vastly different millisecond measurements by submitting the exact same code multiple times, presumably due to noisy neighbors, garbage collection, and other server resource issues. This makes it difficult or impossible to measure the impact of long tail optimizations to the code.
Is there any way to measure the runtime time complexity using actual operations (e.g. frames put on the stack in Java) to be more precise in measurement? Especially for those who are paying for the premium version of Leetcode