SortVivo - Sort Algorithm Visualization

Pingpong merge sort(2014)

O(n log n) ✓ Stable ✗ In-place

🔧 How it works

Recursively splits the array in half like standard Merge sort, but uses two full-size buffers that alternate roles at every recursion level, one buffer is the source (ping) and the other is the destination (pong), swapping at each recursive call.

🎯 Key property

Eliminating the copy-back step of standard Merge sort reduces writes from 1.5n per level to n per level, roughly 33% fewer total writes for large arrays. The result always surfaces in the original array so no final copy-back is needed.

👁 Watch for

RangeCopy the entire input is copied into the auxiliary buffer before recursion begins so both buffers hold valid data; leaf-level copies (dst[i] = src[i]) correctly read from whichever buffer is the current source at that recursion depth
Compare a skip-sorted check (src[mid] ≤ src[mid+1]) fires before each merge; when the two halves are already in order the merge is replaced by a single bulk copy instead of per-element comparisons
IndexWrite during each merge, elements are written directly from the source buffer to the destination buffer without any intermediate scratch space; the buffers swap roles at each recursive level so the sorted result always lands in the original array

🕰 History 🌳 Lineage ⚡ Optimization

UsesPing-Pong Buffering Range Reduction

Initial State

This is the array before sorting. Follow Pingpong merge sort step by step.

Recursion Tree

0 / 112 ops ( 0.0%) 0.0%