<span class="var-sub_title">Revisiting the 2008 ExaScale Computing Study and Venturing Predictions for 2028</span> SC18 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Revisiting the 2008 ExaScale Computing Study and Venturing Predictions for 2028


Authors: Jeffrey Vetter (Oak Ridge National Laboratory), Peter Kogge (University of Notre Dame)

Abstract: On this 10-year anniversary of the ‘ExaScale Computing Study: Technology Challenges in Achieving Exascale Systems,’ we will reexamine the report’s predictions, and use this information to crowdsource projections for HPC architectures in 2028. First, members from the original team, including the study leader, Peter Kogge, will briefly highlight the report’s hits and misses. Second, the panelists will present initial thoughts about important architectural trends for 2028, guided by the outcome from the earlier retrospective. Throughout the session, we will electronically survey the audience on these hits and misses, their predictions for 2028, and then ask the panelists to address them.

Long Description: On this 10-year anniversary of the ‘ExaScale Computing Study: Technology Challenges in Achieving Exascale Systems,’ we will reexamine the report’s predictions, and use this information to crowdsource projections for HPC architectures in 2028. The report’s objectives were to understand the trends of mainstream computing technology, and then determine whether it would allow a 1,000X increase in the capabilities of computing systems by 2015. Moreover, the report enumerated gaps in these technology trends, identified the major challenges, and recommended additional targeted research to provide potential solutions for overcoming these challenges. In the first part of the session, several members from the original team, including the study leader, Peter Kogge, will briefly highlight the report’s hits and misses, and address questions from the audience and appointed inquisitors. In addition, we will electronically survey the audience in real-time on these hits and misses, and then ask the panelist to address them during the discussion. In the second part of this session, the panelists will present initial thoughts about important architectural trends for 2028, guided by the outcome from the earlier retrospective. Once again, we will electronically survey the audience in real-time on their trend predictions and present them during the session prompting discussion and introspection.




Back to Birds of a Feather Archive Listing