On the road to exascale: Advances in High Performance Computing and Simulations - An overview and editorial

In recent decades, the complexity of scientific and engineering problems has increased considerably. New applications and domains that use high performance computing systems have been introduced. These trends are projected to continue for the foreseen future (Reed and Dongarra, 2015) [1]. In many areas of engineering and science, High-Performance Computing (HPC) and Simulations have become determinants of industrial competitiveness and advanced research. In fact, advances in HPC architectures, storages, networking, and software capabilities are leading to a new era in HPC and simulations, along with new challenges both in computing and systems modeling (Geist and Lucas, 2009) [2]. These developments are especially critical considering that HPC systems continue to scale up in terms of nodes, cores, and accelerators, as well as software, infrastructure and tools, which in turn are expediting the move on the path toward Exascale (Reed and Dongarra, 2015; Geist and Lucas, 2009; Dongarra and Beckman, 2011; Dosanjh et al., 2014; Engelmann, 2014) [1–5]. Scalability and availability represent two of the main requirements that need to be considered before conceiving of these large-scale systems (ASCAC Subcommittee on Exascale Computing, 2010). The scalability feature allows the system to proportionately grow when service demand increases, whereas availability means the system continues to provide their services despite hardware and software failures (Theodoropoulos et al., 2014; Tang et al., 2014) [7,8]. The goal in large-scale HPC is to accommodate both availability and scalability while staying under strict constraints on performance (e.g., processing time) and cost metrics (e.g., power consumption). This special issue is envisioned to provide examples of research work on topics related to recent advances in High Performance Computing and Simulations. It briefly addresses and explores challenges toward Exascale computing, current state-of-the-art in HPC and simulation, and the path forward in the domains of large-scale HPC systems.