Milvus 2.3 Massively Accelerates AI-Powered Applications With NVIDIA GPU Support

作者：Zilliz

2023-03-23
上海
本文字数：2592 字
阅读完需：约 9 分钟

NVIDIA GTC 2023—March 21, 2023—Zilliz[1], the inventor of the open-source vector database Milvus, today announced the beta launch of Milvus 2.3, featuring NVIDIA GPU support for greater flexibility and dramatic improvements in real-time workload performance.

Purpose-built for AI-powered applications, Milvus stores, indexes and manages billions of embedding vectors generated by large language models (LLMs), convolutional networks and other machine learning (ML) models. Milvus handles all types of unstructured data, eliminating the cost and complication of managing multiple databases. With Milvus, organizations can quickly analyze and act on their data to reduce fraud, deliver better recommendations, avoid downtime and make faster decisions.

Heterogeneous (CPU + GPU) computing massively improves the performance of real-time recommendation engines, question-and-answer systems, anomaly detection, image and video search and other applications powered by similarity search. Milvus is the first vector database to support heterogeneous compute, and the latest Milvus release brings this powerful capability to the rearchitected cloud-native Milvus 2 platform[2], which offers hybrid search, tunable consistency, always-online operation and many other advantages.

With accelerated parallel performance and quicker, more efficient querying, Milvus 2.3 is 4X faster than Milvus 2.0 and more than 10X faster than competitors that retrofit vector search onto traditional database solutions. Milvus 2.3 also introduces GPU acceleration that delivers 10X faster performance than the CPU-only version. GPU-accelerated Milvus will continue to improve and provide high-performance vector search capabilities for various applications powered by machine learning and AI.

“Heterogeneous compute is the key to delivering the processing performance required for AI-powered applications,” said Charles Xie, creator of the Milvus project and CEO, Zilliz. “With Milvus’s NVIDIA GPU support and RAFT[3]-based integration, that capability is now available at massive scale on CPU and GPU platforms — or both.”

Other notable features of Milvus 2.3 include:

Change Data Capture (CDC)— Captures and delivers a continuous feed of database updates for zero-downtime backup and synchronizationIndex Types — Supports nine different index types for optimal performance, cost-effectiveness and accuracy for individual use cases (FLAT, IVF_FLAT, IVF_SQ8, IVF_PQ, HNSW, ANNOY, BIN_FLAT, and BIN_IVF_FLAT)Range Search — Enables searching for all vectors within a specified distance-Rolling Upgrades — Enables admins to upgrade Milvus with minimal downtime and without noticeable service disruptionOn-Disk Index — Optimizes memory usage.

“Support for NVIDIA GPUs in the latest version of Milvus will bring huge benefits of heterogeneous compute to real-time applications,” said Kari Briski, vice president of software product management at NVIDIA. “Milvus is a highly performant vector database, and with the massive parallelism of NVIDIA GPUs, users can now accelerate compute pipelines.” First released in 2018, Milvus is an LF AI & Data graduated-stage project[4] with more than 200 contributors, 1,000 enterprise users, and 2,000 community members.

For more information on Milvus 2.3, visit https://zilliz.com/blog/milvus-2-3-beta-new-features-and-updates.

About Zilliz Zilliz is a leading vector database company[5] for enterprise-ready AI. Built by the engineers who created Milvus[6], the world’s most popular open-source vector database, Zilliz is on a mission to unleash data insights with AI. The company builds next-generation database technologies to help organizations rapidly create AI/ML applications and unlock the potential of unstructured data. By taking the burden of complex data infrastructure management off of its users, Zilliz is committed to bringing the power of AI to every corporation, every organization, and every individual.

Headquartered in Redwood Shores, CA, Zilliz is backed by a number of prestigious investors, including Aramco’s Prosperity7 Ventures, Temasek’s Pavilion Capital, Hillhouse Capital, 5Y Capital, Yunqi Partners, Trustbridge Partners, and others. Zilliz’s technologies and products help over 1,700 organizations worldwide easily create AI applications in various scenarios, including computer vision, image retrieval, video analysis, NLP, recommendation engines, targeted ads, customized search, smart chatbots, fraud detection, network security, new drug discovery, and much more. Learn more at http://zilliz.com or follow @zilliz_universe.