Explore the fascinating world of sorting networks in this 24-minute talk delivered by Kofi Wilkinson. Delve into the speed-up opportunities provided by sorting networks, which harness the power of instruction-level parallelism. Discover how these networks can improve upon common "optimal" sorting methods like mergesort and quicksort, potentially reducing the O(n log n) time complexity bound of comparison-based routines towards a sub-linear O((log n)^2). Gain insights into this lesser-known but powerful approach to sorting that offers significant performance improvements over traditional methods.