Memory capacity is a major bottleneck in hyperscale datacenters, as approximately 40% of servers are memory-capacity bound, which limits both performance and scalability. Compute Express Link® (CXL®) is emerging as a foundational technology for enabling scalable, composable memory architectures. In this webinar, CXL technology experts from Meta will share insights into Vistara, an end-to-end CXL memory expansion platform spanning custom ASIC design, firmware, operating system enablement, and deployment across millions of production servers. Attendees will learn how CXL is being used to improve infrastructure efficiency, reduce hardware requirements, and support demanding AI and machine learning workloads while addressing common concerns around latency and performance.
Join CXL Consortium member Meta for an in-depth discussion on the lessons learned from deploying CXL at hyperscale, the architectural considerations behind production-ready implementations, and what these advancements mean for the future of memory disaggregation and data center infrastructure. Meta will also highlight deployment results, including up to a 25% reduction in server count for disaggregated machine learning inference and a 29% reduction in average latency for distributed caches, demonstrating the tangible impact of CXL in production environments.
Register for the webinar HERE.