I dare you to try and get heavy CPU cache-level performance optimizations you might see in tried and true HFT code written this way.