3n-sort: Parallel Sorting Algorithm Visualizer

Live app: https://alexandrosliamp.github.io/3n-sort/

An interactive, bilingual (Greek / English) visualizer of 3n-sort, the parallel sorting algorithm by Schnorr and Shamir (1986) that sorts an n×n mesh of processors into snakelike order in 3n + o(n) steps, which is optimal up to lower-order terms.

What it shows

• The full 8-phase algorithm animated on a real grid (16×16 up to 256×256, or any custom size via padding): block sorting, the k-way unshuffle of columns, column and row sorts, and the final odd-even passes along the snake.
• True parallel execution: every compare-exchange pass runs on a pool of Web Workers over a SharedArrayBuffer, mirroring the mesh model instead of simulating it sequentially.
• Sequential vs parallel odd-even transposition sort, raced side by side on the same values: 8 time steps in parallel vs 28 sequentially.
• The unshuffle as dealing cards: n cards dealt round-robin to k players, with the original and resulting rows shown together for comparison.
• Step-by-step phase descriptions, draw-your-own-grid mode, a guided tour, light/dark themes, and a zoomable canvas.

Keywords

parallel sorting algorithm · mesh-connected processor array · 3n-sort · Schnorr–Shamir · odd-even transposition sort · snakelike order · k-way unshuffle · sorting visualization

Running locally

The app needs cross-origin isolation for SharedArrayBuffer. Serve the files over HTTP (a bundled service worker handles the isolation headers on hosts, like GitHub Pages, that don't send them):

python -m http.server 8080
# then open http://localhost:8080/

Files

• index.html — the entire app (single file: HTML + CSS + JS)
• sort-worker.js — the compare-exchange worker kernel
• coi-serviceworker.js — cross-origin-isolation bootstrap for header-less hosts