Fix False Sharing Bug: Two Threads Updating Adjacent Array Elements

Independent Variables, Shared Cache Lines

False sharing happens when two threads write different variables that live in the same 64-byte cache line. Each write forces the line to bounce between cores even though the threads touch disjoint data. OpenMP reductions and per-thread counters often hit this bug.

Detection and Fix

Symptoms: scaling collapses when you add cores to an embarrassingly parallel loop. Fix: pad structs or arrays so each thread's hot field occupies its own cache line.

cCopy
Loading…

• perf c2c on Linux maps line conflicts to source lines
• Align hot counters to 64 bytes with alignas(64)
• Read-only sharing is fine; write sharing on the same line hurts
• Atomics on adjacent counters still false-share without padding

Working Through the Exercise

Keep the relevant documentation open while you implement. When your output disagrees with the reference, trace one failing case by hand before changing random lines.

• Trace first: walk one failing input step by step
• Read the manual: skim the official doc for the mechanism named in this task
• Compare baselines: save measurements from the prior step so speedups are honest

Why for this exercise

You will demonstrate false sharing between two threads updating adjacent counters, then fix it with padding and compare timings. This exercise requires showing improved scaling after separation.