Advanced Computing in the Age of AI | Tuesday, March 19, 2024

A Call for National Action: HPC ‘At the Frontier of Competition’ 

The Information Technology and Innovation Foundation (ITIF), a Washington D.C. think tank with close ties to the Office of Science and Technology Policy and government broadly, today released an expansive report – The Vital Importance of High- Performance Computing to U.S. Competitiveness – and also held a panel to discuss the report’s recommendation. Noteworthy, many of the panelists are familiar names in the HPC community.

All boiled down, the ITIF report is another call for national action in support of HPC, similar in tone to (and supportive of) the National Strategic Computing Initiative (NSCI). Calling HPC a strategic, game- changing technology with tremendous economic competitiveness, science leadership, and national security implications, the ITIF document treads familiar NSCI ground as shown this brief excerpt:

“Because HPC stands at the forefront of scientific discovery and commercial innovation, it is positioned at the frontier of competition—for nations and their enterprises alike—making U.S. strength in producing and adopting HPC central to its competitiveness. But as competitor nations rapidly scale up their investments in and applications of high-performance computing, America will need concerted public and private collaboration and investment to maintain its leading position in both HPC production and application.”

ITIF.KeynoterHow effective this latest call for a coordinated national HPC policy will be is an open question. ITIF’s bipartisan nature may help, say observers, and today’s keynoter before the panel was Republican Randy Hultgren, Representative (R-IL-14) U.S. Congress (shown on the right).

The scheduled panel featured prominent HPC organizations and senior personnel:

  • IntelJoseph Curley, Intel’s senior director, HPC Platform and Ecosystem Enablement in the High Performance Computing Platform Group.
  • Hewlett Packard EnterpriseBill Mannel, vice president and general manager of High-Performance Computing and Big Data.
  • IBM – David Turek, vice president, Exascale Systems.
  • National Renewable Energy Laboratory (NREL) – Steven Hammond, director of the Computational Science Center.
  • IDCRobert Sorensen, IDC research vice president, IDC High Performance Computing.
  • ITIFStephen Ezell, one of the report’s authors and vice president, Global Innovation Policy.

It’s good to recall the draft implementation plan for NSCI has yet to be made public. IDC’s Sorensen, who spent 30-plus years as a senior technology analyst in government said he was among many senior HPC watchers who were extensively interviewed by ITIF for the report, “They were very interested in getting the story right, and their credibility is good. But to be sure, they are not HPC experts, but they have good political cred here in DC.”

Turek of IBM said after the panel, “The report is well intentioned and mostly correct. [It] makes a compelling case that HPC is widely used and particularly beneficial, but who needs to read it to be convinced of it.” Turek wondered how best to handle the ‘operationalizing’ challenge which was not spelled out as clearly as the goals.

Unlike the NSCI executive order, the ITIF report presents a fair number of details and examples of the role and impact of HPC on industry as well a compilation of major global initiatives seeking to obtain HPC leadership. China’s developing plans – including expectations it will fire up two 100 petaflops computers this year – receives lengthy treatment.

ITIF HPC Programs Table

Excerpt: “Clearly, China has made HPC leadership a national priority. A key reason for this is that, for China, leadership in high-performance computing is central to the country’s goal of transitioning away from reliance on foreign technology to using homegrown technology. As Li Na, a spokesperson for the Tianhe-2 project, explains, “We are producing supercomputers with a fundamental purpose of providing a driving force for the construction of an innovation-oriented country.” As IDC’s Rajnish Arora explains, “The Chinese government and companies want to become the creators and not just producer of products that are being designed elsewhere.” Or, as Chinese President Xi Jinping himself puts it, China has built its HPC capabilities in part to demonstrate that the country has become a cyber power.”

In the release announcing the report, ITIF’s Ezell is quoted, “The U.S. is home to three of the five fastest supercomputers in the world, but China is home to the global frontrunner and plans to launch an even faster supercomputer this year. Japan and the EU have also introduced concerted national programs to achieve high-performance computing leadership. While America is still the world leader, other nations are gaining on us, so the U.S. cannot afford to rest on its laurels. It is important for policymakers to build on efforts the Obama administration has undertaken to ensure the U.S. does not get out paced.”

National plans/aspirations in Europe, Japan, Russia, South Korea, and India are also briefly reviewed. For those familiar with HPC, the full report is a fast read – despite its 50-plus-page length.

The ITIF report calls for energizing NSCI and the additional steps listed here:

Congress should:

  • Hold hearings on the National Strategic Computing Initiative (NSCI) and the intensifying race for global HPC leadership.
  • Authorize and appropriate funding levels for the National Strategic Computing Initiative as requested in the Obama administration’s FY 2017 budget for FY 2017 and for future years.
  • Reform export control regulations to match the reality of current high- performance computing systems.

The administration, or its agencies and departments therein, should:

  • Continue to make technology transfer and commercialization activities a priority focus of America’s network of national laboratories.
  • Emphasize HPC in federal worker training and retraining programs.
  • Emphasize HPC in relevant Manufacturing Extension Partnership (MEP) engagements, helping facilitate small- to medium-sized enterprises’ (SME) access to high-performance computing.

Turek made a few suggestions for modification or additions to the report’s policy recommendations. For example, with respect to NSCI, Turek said, “It would be appropriate formally to get leaders from American industry involved to establish the list of industry grand challenges so there is a direct linkage between the activities of NSCI and impact on competitiveness as opposed to thinking it is going to be accomplished through some sort of indirect trickle down effect. I also thought it would be motivating for people to come and work on these problems.”

The expansion of industrial use of HPC is certainly a central tenant of the report. It notes: “Finally, in February 2016, as part of its HPC4Mfg (HPC for Manufacturing) challenge, the Department of Energy announced $3 million in funding for 10 projects that will allow manufacturers to tap into the power of HPC systems at DOE-managed national laboratories.113 Each of the projects is designed to leverage HPC to improve efficiency, enhance product development, or reduce energy consumption. For example, one initiative will help Global Foundries optimize semiconductor transistor design, and in another GE will leverage advanced HPC particle physics simulations to improve the efficiency and lifespan of its aircraft engines.114 The vision is to grow this concept from just HPC4Mfg into an HPC4X template where the same process can be applied to HPC4transportation, HPC4life sciences, etc.”

The compilation of examples of HPC’s impact on industry is extensive and although many are widely known, it’s the scope of HPC’s current potential effect that is most interesting. Here are just three from the report, which is freely available online:

  • “Boeing physically tested 77 prototype wing designs for its 767 aircraft (which was designed in the 1980s), but for its new 787 Dreamliner, only 11 wing designs were physically tested (a 7-fold reduction in the needed amount of prototyping), primarily because over 800,000 hours of supercomputer simulations had drastically reduced the need for physical prototyping.5
  • “HPC has facilitated development of a cloud-based tool that simulates welding processes used in metallic product manufacturing. The application, being developed by the Ohio Supercomputer Center (OSC) and the Engineering Mechanics Corporation of Columbus, in part through a Small Business Innovation Research (SBIR) grant awarded by DOE, is a welding design software package called Virtual Fabrication Technology that enables SMEs to tap into HPC resources, so they can validate the integrity of welds in assembled components.85
  • “Roughly 4,100 genetic diseases affect humans, and they are the main causes of infant deaths. But identifying which genetic disease is affecting a critically ill child is extremely difficult. For one infant suffering from liver failure, the center used 25 hours of supercomputer time to analyze 120 billion nucleotide sequences and narrowed the cause of the illness down to two possible genetic variants. Thanks to this highly accurate diagnosis, the baby is alive and well today.”

One issue raised by Turek is the need to incentivize ISVs. The SME community highlighted in the report relies heavily on commercial codes, but these codes don’t scale well, and many ISVs have established positions of prominence in particular; as a result they have limited motivation to drive advances. Most of the codes were developed in the late 60s and 70s – often with Government funding at national labs – when no anticipated scaling would become the dominant computing paradigm.

Turek made two suggestions, “One, we should reengage ISVs on this issue of modern algorithm development and modern numerical methods in the context of emerging kind of hardware architectures we’re seeing. That would go a long way to helping overcome the scaling issues of today.” Number two, thought needs to be given to how best to “provide incentives to get these commercial ISVS to modernize their software, make it portable, and make it scalable, if there is not a market push to do this.”

If the incentives are there and commercial ISVs “don’t respond well, then redirect that money to the open source community and help them get better established,” said Turek.

Needed training and the potential for expanding the workforce are also covered. The report notes that HPC specifically is an important component of the broader computer manufacturing subsector that in the United States employs approximately 1 million individuals, 600,000 in production roles.34 In 2015, this employment included 28,370 computer hardware engineers; 22,570 semiconductor processors; 38,010 electrical and electronic engineering technicians; and 97,200 electronic and electronic equipment assemblers.

The report is peppered with statistics, quotes, and examples and is best read in full to form one’s own conclusion. It will be interesting to see if the ITIF report can generate real activity. Sorensen, who participated in the early development of NSCI noted, “I am really looking to see if this is a chance to get the Hill engaged.”

You can read the full report here and watch the video of the panel here.

EnterpriseAI