Designing Development Readiness Indicators and Roadmap for Petaflops Scale Supercomputers

Recently, supercomputers and high-performance computing (HPC) have caused significant progress in many fields, including industrial and research fields of artificial intelligence, data science, big data, Internet of Things, as well as large scale web applications. Regarding this characteristic, the supercomputer field becomes one of the strategic issues in developed and developing countries. Successful HPC development requires a robust infrastructure that effectively depends on a multidimensional policy and roadmap, covering all aspects of HPC services. In this regard, a comprehensive roadmap is necessary to design and achieve high-performance computing services. This paper proposes a comprehensive roadmap focusing on several strategic areas, including research activities, infrastructure and platforms, data processing, standard regulations, applied services, and business development. The roadmap follows an evolutionary strategy where strategic regions evolve in parallel. Also, in this paper, the Supercomputing Development Readiness Indicators (SDRI) inspired by the "Network Readiness Indicators" (NRI) are presented and the proposed roadmap uses them. The proposed roadmap is then compared with two other hypothetical roadmaps with the timing and priority of different strategic areas in terms of progress and completion time. Evaluation of the proposed roadmap based on the SDRI specification shows the effectiveness of the proposed method in creating and developing the services of a petaflops scale supercomputer.

Keywords:

HPC Services , roadmap , Strategic domain , Coevolutionary