This talk focuses on enabling GPU accelerated vector search in Apache Solr using Lucene-cuVS connector and Nvidia’s cuVS library - A library that provides implementations of several algorithms for approximate nearest neighbor and clustering on GPU.

This talk will have three main sections:

[1] Apache Solr, its introduction, current vector search implementation, and enabling accelerated vector search using Lucene-cuVS connector and its comparison.

[2] Lucene-cuVS and its architecture, challenges, and benchmarking. How Lucene-cuVS enables accelerated vector search in large workloads.

[3] Introduction to Nvidia’s cuVS library, its history, ANN search algorithm types, their implementations and comparison, the novel graph-based CAGRA algorithm, and the cuVS future roadmap.

Speakers:
[1] Corey J. Nolet - Senior Data Scientist & Software Architect, Nvidia
[2] Ishan Chattopadhyaya - Apache Lucene & Apache Solr Committer, SearchScale
[3] Vivek Narang - Senior Software Engineer, SearchScale