Persistent Memory and the Path to Being Comfortably NUMB

Speaker: Steven Swanson (UCSD)

Abstract: Memory systems are on the verge of a renaissance:  Scalable, persistent main memories (e.g., Intel’s 3DXPoint) are the first new technology to enter the upper layers of the memory hierarchy in 50 years.  They bring a fundamentally new capability (i.e., persistence), a dramatic increase in capacity, and an array of complications (e.g., asymmetric read and write performance, power limitations, and wear out).  This combination of characteristics raises a deceptively simple but fundamental question:  What should we do with persistent main memory?  In this talk, I will describe several potential answers and the systems  my group has built to help understand how different answers affect performance, programmability, and other aspects of system design.   I’ll also highlight the central challenges that these memories present and try to summarize what we have learned about them.  Finally, I’ll describe what I see as the most interesting avenues for future work.

Bio: Steven Swanson is a professor in the Department of Computer Science and Engineering at the University of California, San Diego and the director of the Non-volatile Systems Laboratory.  His research interests include the systems, architecture, security, and reliability issues surrounding heterogeneous memory/storage systems, especially those that incorporate non-volatile, solid-state memories.  He has received an NSF CAREER Award, Google Faculty Awards,  a Facebook Faculty Award, and been a NetApp Faculty Fellow.  He is a co-founder of the Non-Volatile Memories Workshop. In previous lives, he worked on low-power co-processors for irregular applications and building scalable dataflow architectures. He received his Ph.D. from the University of Washington in 2006 and his undergraduate degree from the University of Puget Sound in 1999.