For the last 60 years, high-performance computing (HPC) has been instrumental in chipping away at the world’s toughest challenges such as disease control, climate research, and energy efficiency. Organizations in many industries, including oil and gas, financial services, pharmaceutical, and life sciences, as well as in academia and government, have drawn on this technology to make scientific advances, create innovation, and understand complex data. Because the cost of designing and building new generations of HPC systems is high, the US Government has always been involved—in close collaboration with industry—as a major funding source, as well as a major HPC user.
However, with a July 2015 White House executive order, the US Government lit a fire under US HPC research in a big way by creating the National Strategic Computing Initiative (NSCI). Why is the Federal Government amping up its commitment to HPC now? The timing is purposeful. Throughout the government, there has been broad acknowledgement that HPC has contributed substantially to national economic prosperity, accelerated scientific discovery, and national security, and agreement that the US cannot afford to lose its international competitive edge. This executive order lays the groundwork to achieve productive exascale capabilities in the US by the target timeframe of 2023.
What is the NSCI, and why is it a big deal?
The goal of the NSCI is to ensure that the United States continues to lead in HPC over the coming decades. The NSCI will draw on the strengths of multiple US Government departments and agencies to move the Federal government into a position that sharpens, develops, and streamlines a wide range of new 21st century applications. It is designed to advance core technologies to solve difficult computational problems and foster increased use of the new capabilities in the public and private sectors.
The NSCI has five strategic objectives:
- Accelerate delivery of a capable exascale computing system that integrates hardware and software capability to deliver approximately 100 times the performance of current 10-petaflop systems across a range of applications representing government needs.
- Increase coherence between the technology base used for modeling and simulation, and that used for data analytic computing.
- Establish, over the next 15 years, a viable path forward for future HPC systems even after the limits of current semiconductor technology are reached.
- Increase the capacity and capability of an enduring national HPC ecosystem by employing a holistic approach that addresses relevant factors such as networking technology, workflow, downward scaling, foundational algorithms and software, accessibility, and workforce development.
- Develop an enduring public-private collaboration to ensure that the benefits of the research and development advances are, to the greatest extent, shared between the United States Government and industrial and academic sectors.
The NSCI is significant for several reasons. First, it’s an all-government approach to HPC. Whereas previous government funding efforts were siloed through and splintered across specific agencies, NSCI presents a single, coherent strategy that all US Government agencies will get behind. It’s the first time in many years that HPC excellence is being set up as a cross-agency national goal—an Apollo project, if you will.
The NSCI is also significant in its charter to merge HPC and high-performance data analytics (HPDA), which heretofore have been on different tracks. (See objective #2 above.) HPC has been focused on applying massive levels of computational power (petaflops) to theoretical problems in academic institutions and supercomputing centers. HPDA, the new name for Big Data, has been concerned with analyzing massive amounts of data (petabytes), primarily to aid business decision-making. Data-intensive HPDA environments tend to look very different from their computing-intensive counterparts; the HPDA systems tend to be comprised of off-the-shelf commodity components rather than customized purpose-built systems.
Through expanded program initiatives by designated federal agencies already investing in HPC, the NSCI seeks to merge these two disciplines and revolutionize our problem-solving capabilities, combining the best attributes of today’s compute-intensive and data-intensive architectures. It seeks to produce systems that can perform complex modeling and simulation to derive insightful theoretical outcomes while at the same time respond to massive volumes of real-time information. The convergence of petaflops and petabytes will bring supercomputing to the masses, enabling more private organizations to participate in solving the world’s biggest challenges.
How Intel is involved
As the maker of the microprocessors and components that power many of the world’s supercomputers and HPDA storage systems, Intel is excited about the NSCI and is in a unique position to help advance all five NSCI objectives. We’ve established an Extreme Scale Program office in Washington, D.C., and we were involved during the two-plus years of research and policy formulation that led up to the NSCI.
Intel pours millions of dollars annually into developing new microprocessor technologies. We are investing not only in Complementary metal–oxide–semiconductor technologies but in quantum, neuromorphic, and other technologies that will be needed to take HPC to exascale and beyond.
As the industry fixes its sights on exascale, some have sounded the death knell for Moore’s Law, which has guided and governed the modern electronics industry for 50 years. However, Intel is working hard to sustain it. In fact, we believe that the path to exascale depends on it. The new Intel Scalable System Framework (Intel SSF)—an advanced approach for developing scalable, balanced, and efficient HPC systems—was designed with Moore’s Law in mind. Intel SSF takes advantage of Intel innovations such as the Intel Omni-Path Architecture fabric and 3D XPoint Technology (a new class of non-volatile memory) to power the next generation of supercomputers. We believe that what many have labeled as the “post-Moore’s Law” era is actually quite heavily influenced still by Moore’s Law.
Through its Extreme Scale Program office, Intel is already at work in several public-private partnerships. One of our key focus areas is with the US Department of Energy on the CORAL program. In November 2014, the Department of Energy awarded $325 million to build state-of-the-art supercomputers at its Oak Ridge, Argonne, and Lawrence Livermore National Laboratories. The joint Collaboration of Oak Ridge, Argonne, and Lawrence Livermore (CORAL) was established in early 2014 to leverage supercomputing investments, streamline procurement processes, and reduce costs to develop supercomputers that will be five to seven times more powerful when fully deployed than today’s fastest systems in the US.
Argonne National Laboratory chose Intel and its integrator partner Cray to build its next-generation systems, the first of which (Theta) comes online in late 2016 and the second of which (Aurora, an 180-Petaflops system that will likely be the world’s most powerful supercomputer when installed, will be delivered in 2018. They are both Intel Xeon Phi processor-based systems with deep memory architectures.
Intel and its partners are also working closely with several US Government agencies on code optimization, training, and application development in many areas, including precision medicine, weather forecasting, security, and defense.
Intel is working with the three lead NSCI agencies—Department of Energy, Department of Defense, and National Science Foundation—named by the NSCI to develop and deliver exascale-based HPC systems. Our work with the CORAL systems falls in this category.
We’re also working with the two research and development agencies—the Intelligence Advanced Research Projects Activity (IARPA) and the National Institute of Standards and Technology (NIST)—which are charged with fundamental scientific discovery work and associated engineering advances needed to support the NSCI objectives.
Lastly, we’re working with the five NSCI deployment agencies— the National Aeronautics and Space Administration, the Federal Bureau of Investigation, the National Institutes of Health, the Department of Homeland Security, and the National Oceanic and Atmospheric Administration—most of which are charged with developing mission-based HPC requirements that power requirements of government-employed principal investigators and, in some cases, drive collaboration with the private sector and academia
It’s an exciting time for HPC, and Intel is thrilled and proud to be involved at such a level of collaboration and technology leadership on an initiative as important as the NSCI. Exascale computing suddenly seems closer at hand than ever.