MEMOCODE 2016 design contest: K-means clustering

Abstract

K-means is a clustering algorithm that aims to group data into k similar clusters. The objective of the 2016 MEMOCODE Design Contest is to implement a system to efficiently partition a large set of multidimensional data using k-means. Contestants were given one month to develop a system to perform this operation, aiming to maximize performance or cost-adjusted performance. Teams were encouraged to consider a variety of computational targets including CPUs, FPGAs, and GPGPUs. The winning team, which was invited to contribute a paper describing their techniques, combined careful algorithmic and implementation optimizations using CPUs and GPUs.