Maze Race — pathfinding algorithms, head to head

start goal solution dim → bright = order explored

Results

AlgorithmCells exploredPeak frontierPath length OptimalSolve timeFinish
Cells explored is how much of the maze the search had to look at. Peak frontier is how many cells it held in memory at once. Solve time is real CPU time, measured over repeated un-instrumented runs (best of N) — which is why Dijkstra can be slower than BFS while exploring exactly the same cells: it pays for a priority queue. Every search runs to completion the instant you press Race — the animation is a recorded replay, advancing each algorithm at the same cells-per-frame, so what you watch is explored-cells, not wall-clock time. Times are single-threaded JavaScript in this browser; compiled C on the same maze would be several times quicker again.