Persistent Ferroelectric Memories for Artificial Intelligence Compute

Abstract: Dynamic random-access memory plus (DRAM+) is a novel nonvolatile ferroelectric memory technology that delivers DRAM‑class latency and bandwidth while providing inherent data persistence, effectively bridging the long‑standing gap between volatile main memory and storage.

By eliminating refresh operations, DRAM+ significantly reduces power consumption and thermal load, enabling energy‑efficient system designs and instant‑on capability. Crucially, the technology is manufactured using processes that are more than 99% identical to conventional DRAM, allowing seamless integration into existing DRAM fabs with minimal capital investment and clear cost and scalability advantages over other emerging memories.

This unique combination of performance, persistence, manufacturability and endurance simplifies system architectures and makes DRAM+ particularly well suited for artificial intelligence, high‑performance and data‑centric computing workloads that demand low latency, high bandwidth and persistent memory at scale. 3D heterogeneous integration further enhances the flexibility and applicability of such memories.

Series: The Argonne Microelectronics Institute colloquium series invites leaders in the field from academia, national labs and industry to present their forefront research and perspectives. Join us in-person or online to learn and connect.

If you have questions about the colloquium series or would like to suggest or host a future speaker, contact Dan Durham (durhamd@​anl.​gov).