Here is the abstract you requested from the thermal_2013 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|The Future of Computer Architecture and Implications on Thermal Management|
|Keywords: thermal management, computer architecture, design|
|The way in which computer systems are designed and built has gone largely unchanged over the past 50-plus years. While components have gotten smaller and more integrated, enabling significant performance and efficiency improvements, the basic architecture of machines continues to be based on traditional compute, memory and storage hierarchies that use copper wires for communication. This design paradigm is in the process of undergoing several 'tectonic shifts' that will change the way in which computer systems are assembled and programmed, and which will enable new ways to interact with the explosion of data that is expected over the next decade. This talk will explore three fundamental technologies that are expected to converge and provide the foundation for the next generation of system architecture within the next decade. In particular, it will explore the emergence of non-volatile memory and its role in the flattening of the traditional storage-memory hierarchy. Evolutionary changes in microprocessor design, particularly with respect to chip-level integration will also be discussed. Additionally, the role of photonics as a potential replacement for copper wires in communication fabrics will be examined. Finally, for each technology shift, the impact on thermal architecture and energy management will be explored and challenges and opportunities will be highlighted.|
|Cullen Bash, Research Manager
Palo Alto, CA