VStore: In-Storage Graph Based Vector Search Accelerator
TimeThursday, July 14th11:37am - 12pm PDT
Location3002, Level 3
Event Type
Research Manuscript
In-memory and Near-memory Computing
DescriptionGraph-based vector search that finds best matches to user queries based on their semantic similarities using a graph data structure, becomes prevalent in data science and AI application. However, deploying graph-based vector search engines in production systems requires high accuracy and efficiency with low latency and memory footprint, which existing works fail to offer. We present VStore, a graph-based vector search solution that collaboratively optimizes accuracy, latency, memory, and data movement on large-scale vector data based on in-storage computing. Our evaluation shows that VStore exhibits significant throughput improvement and energy reduction while attaining accuracy over CPU, GPU, and ZipNN platforms.