Research Article
Damian Dechev and Gilbert H
Abstract
The next decade will see a rapid evolution of HPC node architectures as power and cooling constraints are limiting increases in microprocessor clock speeds and constraining data movement. Future and current HPC applications will have to change and adapt as node architectures evolve. The application of advanced exascale architecture simulators will play a crucial role for the design and optimization of future data intensive applications. In this paper, we present our imulation-based framework for analyzing the scalability and performance of massive interconnected networks.