Why it matters
Sorting networks enable parallel + hardware. Understanding shapes parallel.
Advertisement
The architecture
Sequence of comparators (compare + swap).
Fixed structure irrespective of input.
Depth vs size trade-off.
Advertisement
How it works end to end
Bitonic: O(n log^2 n) size, O(log^2 n) depth.
Batcher's merge-exchange.
AKS O(n log n) theoretical.
SIMD implementations.