https://monoid.github.io/tags/atomic/Recent content in Atomic on Monoidal TranformationHugo -- 0.147.2en-usCopyright © 2025, Ivan Boldyrev; all rights reserved.Sat, 27 Jun 2026 14:29:28 +0200https://monoid.github.io/posts/false-sharing-alignment/Sat, 27 Jun 2026 14:29:28 +0200https://monoid.github.io/posts/false-sharing-alignment/<h1 id="what-is-false-sharing">What is false sharing</h1> <p>What is <a href="https://en.wikipedia.org/wiki/False_sharing">false sharing</a>? When CPUs and their cores read and update atomic variables, special hardware protocols make it correct and efficient, while keeping each core&rsquo;s caches consistent. The coordination doesn&rsquo;t happen per-address: these protocols work on cache-line-sized chunks.</p> <p>What happens when two atomic variables happen to reside on the same cache line? For example:</p> <ul> <li>an array of atomics where each atomic is used by a separate thread (e.g. per-thread counters);</li> <li>a queue with two atomic pointers where a writer updates the tail and a reader updates the head. Of course, you declare both pointers in the same struct and they have adjacent addresses!</li> </ul> <p>Each update operation makes the cache line dirty and requires coordination between CPUs. The CPUs are essentially playing ping pong with the chunk of memory.</p>