Unleashing Similarity Search with Intel Optane Persistent Memory

Similarity search is the backbone of many modern applications, such as web search, media (e.g., image, video, or audio) search, and recommender systems. With exploding datasets, Similarity Search or Approximate Nearest Neighbor (ANN) Search at extreme scales is bottlenecked by existing hardware in performance and cost. Join Intel Principal Engineer, Dr. Jawad Khan, as he describes Intel’s winning entry in the NeurIPS Conference Competition “Big ANN Challenge.” He’ll describe the winning algorithm, design choices, the Intel Xeon + Optane PMem hardware set-up and results – 4X better CAPEX+OPEX than the alternative options. Presented by Dr. Jawad Khan, Principal Engineer, Intel Corporation.