Paper Title: Disaggregated Memory Nodes Using a Memory-Centric Software Architecture

Paper Abstract: Disaggregated Memory in the data center will solve problems of memory cost and capacity, but will introduce new challenges due to its remoteness and latency. This talk will discuss the new software foundations required for addressing these new challenges in the context of data-intensive applications such as analytics and deep learning. Our architecture represents a shift from memory as a device to memory as a service. We will illustrate this principle using Elephance's software which is designed to run on disaggregated memory nodes, and which targets the optimization of both data-intensive computation and communication on CPUs and GPUs alike.

Paper Author: , ,

Author Bio: