MEMOCODE 2014 design contest: k-Nearest Neighbors with Mahalanobis distance metric

The MEMOCODE 2014 hardware/software codesign contest problem is k-Nearest Neighbor search using the Mahalanobis distance metric. Given a data set of points in multi-dimensional space, the goal is to find the k points that are nearest to any given point in that space (quantified with the given distance metric). Contestants were given one month to develop a system to perform the kNN search, aiming to maximize performance or cost-adjusted performance. The two winning teams, which have been invited to contribute papers describing their techniques, combined algorithmic and implementation optimizations. The pure-performance winners targeted the Convey HC-2ex hybrid FPGA/multicore system, while the winners for cost-adjusted performance targeted an Intel multicore.