On the Inevitability of Integrated HPC Systems and Their Impact on the Software Stack

High-Performance Computing (HPC) is at an inflection point in its evolution. General-purpose architectures approach limits in terms of speed and power/energy, requiring the development of specialized architectures to deliver accelerated performance. At the same time, data movement has been identified as a main culprit of energy waste, pushing hardware designers towards a tighter integration of the different technologies. The result is a trend to integrated systems, which offer great opportunities in terms of power/performance tradeoffs, but also lead to challenges on the software side. They require new concepts in malleability in operating systems and programming models, all the way to system-wide resource management. We are attacking these challenges in two EuroHPC projects: DEEP-SEA, covering a comprehensive software stack for the first European exascale systems including new additions to MPI to support malleable applications, and REGALE, providing adaptive resource management across workflows, with a special focus on power and energy management. Combined, they will address these challenges with that introduce a novel way to design, program, and operate HPC systems.