Sorting Algorithms Under the Hood
Sorting is measured in comparisons on paper and in cache misses in production. This video traces each classic algorithm at the instruction level: what a swap costs, when the branch predictor wins, and why real libraries ship hybrids instead of pure sorts.
Topics covered:
- What a comparison costs versus a swap or a move
- Quicksort's partitioning and the worst-case pivot problem
- Merge sort's allocation overhead and its stable ordering
- Insertion sort's cache-friendliness on nearly-sorted input
- TimSort and why production libraries use hybrid strategies
- Bucket and counting sort: when sorting is O(n) because it skips comparisons
Related articles
Big-O Measures Scaling, Not Speed
O(n) says nothing about how fast your code is today — it predicts how it behaves as the input grows. What complexity analysis is really for, and when it misleads.
Hash Tables Aren't Magic
Load factors, collision strategies, and cache misses — why O(1) has a constant that bites when you least expect it.
Amortized Analysis: When O(1) Is Really O(1)
Master aggregate, accounting, and potential methods to prove that worst-case-per-operation bounds hold even when individual operations are expensive.
More in Algorithms
Hash Tables Under the Hood
Load factors, collision strategies, and cache behavior — why O(1) has a constant that can be 1ns or 200ns.
WatchTries and String Search
Tries, radix trees, and Aho-Corasick — how prefix structures make lookups proportional to key length, not dataset size.
DetailsHeaps and Priority Queues
Binary heaps as arrays — sift up, sift down, and why insert and extract-min are both O(log n) with no pointers.
DetailsHash Maps and Collision Resolution
Hash maps from the memory up — hash functions, collision strategies, and the load factor that decides when a map resizes.
DetailsGraph Algorithms: BFS and DFS
BFS and DFS traced node by node — the queue versus the stack, the visited set, and what each traversal order is good for.
DetailsDynamic Programming, Visualized
Memoization and tabulation as data flow — overlapping subproblems, the DP table, and how state transitions map to code.
DetailsBinary Search Explained
Binary search at the instruction level — midpoint math, cache behavior, and why it beats linear scan on sorted contiguous data.
DetailsRecursion and the Call Stack
How recursive functions actually execute — stack frames, the base case, and what happens to memory when recursion goes deep.
DetailsBig-O Notation, Visually Explained
What O(1), O(n), and O(log n) mean as instruction counts — growth curves drawn to scale, constants, and real workloads.
DetailsDepth, delivered weekly
One technical dispatch a week — articles and episode notes before they go public.
One technical dispatch per week. No noise.