Accelerating Data Analytics near Memory: A k-NN Search Case Study
TimeTuesday, July 12th6pm - 7pm PDT
LocationLevel 2 Lobby
Event Type
Networking Reception
Work-in-Progress Poster
DescriptionIn this paper, we propose a Accelerating Data Analytics near Memory (ADAM) applying near data processing approach to solve memory side bottleneck. Our analysis shows that the k-NN search is memory-intensive workload, which is suitable for ADAM to process. Based on this analysis, we propose a system equipped with ADAM cards that accelerates k-NN search without the analytics servers involved in the acceleration. Simulation results on various feature vector data sizes show 56.7%, 67.3%, and 68.5% improvements in performance, power, and cost, respectively, compared to baseline system.