Improvements to OpenSearch using Intel Sapphire Rapids and hardware accelerators

In this session, we will share performance improvements in vector search scenarios due to the use of Intel Sapphire Rapids on AWS. The value of hardware accelerators such as AVX512 in Lucene and Faiss libraries will be discussed, along with performance results. We will further dive into the topic of compression, and some innovations in that space. This will focus on using Intel QAT (Quick Assist Technology), which is a compression accelerator present in Intel Gen4 Sapphire Rapids and newer architectures. This accelerator will highlight how hardware acceleration can lower your TCO, and datacenter power efficiency metrics.