DMC

The Data-Model Convergence (DMC) Initiative is an ambitious five-year effort to create the next generation of scientific computing capability through a multidisciplinary software and hardware co-design methodology.

DMC Homepage

Introduction

Our multidisciplinary team is combining scientific modeling and simulation computing paradigms with data analytics to accelerate scientific discovery and enable progress in design, management, and control of large-scale engineered systems such as the power grid.

Our goal is to take current approaches for scientific high-performance computing, deep learning, and graph analytics computing paradigms and integrate them into a converged, coherent computing capability. The DMC Initiative is tackling challenge problems that benefit from the convergence of these three computing paradigms, support the development of a converged programming model and associated runtime system that maps the heterogeneous application workload to DMC’s conceptual heterogeneous processor design. DMC will help position PNNL to define the “Beyond Moore’s Law” computing world, showing that computing capability can indeed be expanded to accelerate scientific discovery well into the 21st century.