Demonstrate False Sharing and Cache Coherency Overhead

False sharing is a real, silently catastrophic bug: Java added the @Contended annotation and the LMAX Disruptor pattern pads its ring-buffer cursors specifically to prevent it, and the Linux kernel's per-CPU variables are deliberately cache-line-aligned for the same reason. Any multi-threaded counter or lock-free data structure that ignores this, from database engines to game engines, can silently lose 10-20x throughput exactly as you measure here.