The "National Strategic Computing Initiative Strategic Plan" was issued last week. The plan wants to accomplish five strategic objectives in a U.S. government collaboration with industry and academia: (1) accelerate the successful deployment and application of capable exascale computing; (2) ensure that new technologies support coherence in data analytics as well as simulation and modelling; (3) explore and accelerate new paths for future computing architectures and technologies, including digital computing and alternative computing paradigms; (4) holistically expand capabilities and capacity of a robust and enduring HPC ecosystem; and (5) establish an enduring public-private collaboration to ensure shared benefit across government, academia, and industry.
In detail the five objectives want to achieve:
Objective 1: The NSCI seeks to accelerate the development of HPC systems with the goal of achieving capable exascale computing by the mid-2020s to meet the needs of critical government applications and capitalize on emerging datasets. Although custom HPC systems that achieve peak performance of an Exaflop/s on specialized benchmarks could be realized by the end of this decade, these systems would not be able to sustain performance at that level on realistic calculations, be expensive to power, be very difficult to program, lack the memory capacity to perform current and anticipated new classes of scientific and engineering applications, and be potentially unreliable in performing long calculations. Exascale computing systems with important characteristics such as affordable power consumption, programmability, reliability, and adequate memory and networking are not projected to become available commercially until the mid to late 2020s, according to industry roadmaps, the Plan says.
Objective 2: Here, the NSCI seeks to develop a coherent platform for modelling, simulation, and data analytics. This should be done primarily through the development of a more agile and reusable HPC software portfolio. Historically, there has been a separation between data analytic computing and modelling and simulation. Systems have been optimized for a specific class of applications, but the differences between these application spaces are fading rapidly.
Objective 3: To develop new computing capabilities, the NSCI will pursue two parallel lines of effort over a 10-20 year period: (1) the research and development of technologies that will move digital computing performance past the theoretical limits of complementary metal-oxide semiconductors (CMOS), and (2) the development of alternative paradigms that will open up new possibilities for large-scale computing. It would be premature to select individual technology paths for alternative computing paradigms or alternative digital computing technologies already, so the NSCI will initially focus on broad research to identify areas of promise across the span of future computing technologies.
Objective 4: The NSCI seeks to develop and adopt new approaches, technologies, and software architectures to support HPC application development, reusability, trustworthiness, sustainability, and HPC workforce development. NSCI also seeks to increase access to HPC technologies and platforms. The current HPC ecosystem of software, hardware, networks, and workforce is neither widely available nor sufficiently flexible to support emerging opportunities, the NCSCI says. The need for change is driven by the growing complexity and size of a broad array of simulations; the expansion of the role of large-scale computation in emerging frontiers of science; and the need for dynamic interaction of computation with other elements of the cyberinfrastructure, including distributed participants, instruments, large data repositories, high-bandwidth networks, and mobile devices.
Objective 5: The NSCI also seeks to broaden public-private collaboration for advancing HPC technologies to support Federal missions, scientific discovery, and economic competitiveness. From the vector-based supercomputer era to the massively parallel systems of modern HPC, collaboration between U.S. Federal agencies and the private sector drove the transformations that define today's international HPC landscape. The inclusive nature of this public-private collaboration endeavors to benefit the public and private sectors, across missions for scientific discovery, economic competitiveness, and national security.
According to the plan, realizing the vision of the NSCI will demand a fully developed HPC ecosystem that meets the needs of government, industry, and academia. Some aspects of that ecosystem have broad commercial drivers and will naturally evolve from the current semiconductor environment. The Strategic Plan (Plan) focuses on areas where government engagement is essential in creating the technological capability, computational foundations, and workforce capacity to realize the vision of the NSCI. The Strategic Plan also identifies the roles assigned to Federal agencies, and highlights ongoing and planned activities that will contribute to NSCI's goals.
The success of the NSCI depends upon development, commercialization, and deployment of new HPC technologies and infrastructure. The NSCI wants to establish collaboration among the U.S. Federal Government, industry, and academia to develop the technological capabilities and computational foundations that will support scientific discovery and economic drivers for the 21st century, and that will not naturally evolve from current commercial activity, the plan says.
The document with the Strategic plan is available at National Strategic Computing Initiative Strategic Plan