The convergence of HPC and embedded systems in our heterogeneous computing future

Recently we have seen two exciting trends that have been flooding the market: 1) the movement of graphics processing units into mainstream general-purpose platforms, and 2) the movement of multi-core embedded systems into tablet computing and smartphone spaces. These trends are forcing application developers to rethink how they are going to best utilize these many-core and multi-core heterogeneous platforms to provide new levels of cost/performance/power in a range of emerging application domains. A key driver in this movement is the recognition that traditional graphics devices can play a larger role in computation than was ever considered before. The high performance computing community has fueled this development, and has demonstrated that Graphics Processing Units can be utilized in a range of general purpose and embedded systems applications. By making this class of devices programmable, a new era in heterogeneous computing has begun. We will discuss some of the catalysts behind these changes, and consider what lies ahead in the future for heterogeneous computing. We will touch on current trends in computing core architectures and programming frameworks, as well as discuss what new classes of applications will be possible as we arrive at the intersection of these two vastly different computing domains.