Intel Accelerators on Amazon OpenSearch Service improve price-performance on vector search by up to 51%

AWS and Intel have announced performance improvements for vector search on Amazon OpenSearch Service using 4th generation Intel Xeon instances with AVX-512 technology.

• Up to 51% improvement in vector search performance on OpenSearch 2.17+ domains
• Acceleration applies to C/M/R 7i instances compared to previous R5 instances
• Enables faster and more cost-effective vector search for generative AI applications
• Intel AVX-512 provides data parallel processing and reduced instruction path length
• Benchmarks show significant gains in both Lucene and FAISS vector search engines

The improvements offer developers a more efficient way to implement vector search, particularly for applications like recommendation systems and AI-powered search experiences.