Find order.
Watch the thinking.
One array. Twenty-two approaches. Make every operation visible.
to clarity.Compare ideas, not just animations.
Two algorithms. Separate measurements.
Race starts both animations together.
| Algorithm | Comparisons | Swaps | Main-array writes | Compute | Animation | Extra space estimate | Average time | Stable |
|---|
Read the experiment
Compute time measures actual sorting plus trace recording in a background worker; it excludes drawing and animation. It is a local, instrumented measurement, not a production benchmark. Animation time counts active playback only. Finishing an animation first does not mean an algorithm is faster in practice.
Comparisons count ordering tests between keys. Main-array writes count overwritten output slots (two per swap); auxiliary-buffer writes, validation, and distribution scans are excluded. N/A means the selected implementation does not use that operation. Progress measures recorded operations played, not a claim that a percentage of positions is final. Green marks a verified completed array.
Space estimates describe algorithm storage, excluding input and visualization. KiB estimates assume 8 bytes per logical slot; JavaScript object overhead and runtime allocations are excluded, so these are not measured heap usage. The trace holds at most 180,000 events per panel; workers have a 1.5-second computational budget and a 5-second watchdog. A capped algorithm can stop without sorting. Categories describe typical complexity, never a universal speed ranking. n = items, k = range/buckets, d = digits, b = radix (10).