Scalable Parallelism in the Extreme | 19 505

Opportunity Information: Apply for 19 505

The National Science Foundation grant opportunity titled Scalable Parallelism in the Extreme (SPX) focuses on the growing reality that computing performance can no longer rely on steady gains from faster single processors. Over the past two decades, computing shifted from mostly single-core chips to systems dominated by multicore and many-core processors, plus large-scale cloud and warehouse-style computing. At the same time, semiconductor scaling is running into physical and practical limits, and single-thread performance has largely leveled off. In plain terms, the old strategy of waiting for the next generation of chips to make software faster is not working anymore. SPX is positioned as a research response to that problem, aiming to push performance forward by making parallelism not just common, but truly scalable at extreme levels.

The opportunity is explicitly tied to the National Strategic Computing Initiative (NSCI), which calls for collaborative work to chart a viable path for high-performance computing over roughly the next 15 years in what is often described as the post-Moore's Law era. SPX treats parallelism as one of the most promising ways to keep improving computational capability, but it emphasizes that the next generation of systems will operate at scales where small inefficiencies become major barriers. For example, short serial portions of code that might be tolerable today can severely limit speedup in massively parallel settings (a classic parallel scaling problem). Likewise, the increasing mix of heterogeneous hardware such as CPUs, GPUs, and specialized accelerators creates difficult challenges in how software schedules work, moves data, and manages different memory hierarchies efficiently.

SPX highlights several concrete issues that become more severe at extreme scale. One is that even tiny non-parallel regions can dominate runtime and undermine the benefits of large numbers of processing elements. Another is heterogeneity: modern systems are not uniform, and coordinating computation across different kinds of processors with different performance characteristics and memory behaviors is complicated. A third challenge is reliability and latency: as systems become more complex, the likelihood of component failures and the cost of communication and coordination can rise, potentially leading to unacceptable delays or reduced mean time between failures even when individual parts are reliable. A fourth challenge is performance predictability in large distributed architectures, where the interconnected nature of the system makes it hard to understand, model, and guarantee performance behavior under real workloads. These examples are presented as representative, not exhaustive, of the broader set of scaling barriers that SPX wants researchers to tackle.

The program aims to fund research that addresses these challenges end-to-end, requiring collaboration across multiple layers of the computing stack. That means work could span from higher-level services and applications, through programming models, runtimes, and system software, down to compilers, operating systems, and even micro-architectural ideas. The intent is not just incremental tuning, but foundational advances that make parallel computing more programmable, efficient, and dependable as scale increases. The solicitation also encourages coordination with industry efforts that share similar goals, reflecting the reality that many advances in extreme-scale computing depend on alignment between academic research, commercial hardware and software ecosystems, and national computing infrastructure priorities.

In terms of how NSF frames the broader outcomes, SPX is meant to align with all five NSCI strategic objectives. While the solicitation text summarizes these at a high level, the theme is consistent: develop new architecture and software approaches that can deliver performance gains after traditional chip scaling slows; support programmable, scalable, and reusable platforms that can be deployed across national HPC and scientific cyberinfrastructure; better connect data analytics computing with modeling and simulation workflows; and advance the capability of extreme-scale computing overall. In other words, SPX is not only about making parallel programs faster, but also about making future HPC systems and software ecosystems more usable, more scalable, and more coherent across different styles of scientific and data-driven computing.

Key administrative details from the opportunity listing include that it is an NSF discretionary grant in the Science and Technology and other Research and Development activity category, under CFDA 47.070. The funding opportunity number is 19-505. The posting indicates an expected 10 awards, and the award ceiling is listed as 0 (which typically signals that proposers must consult the full solicitation for budget guidance rather than relying on a single cap in the summary metadata). The original closing date shown is January 17, 2019, and the opportunity creation date is October 19, 2018. Eligibility is summarized as "Others (see text field entitled Additional Information on Eligibility for clarification)," meaning the full solicitation would need to be checked to confirm which organizations and project types qualify.

• The National Science Foundation in the science and technology and other research and development sector is offering a public funding opportunity titled "Scalable Parallelism in the Extreme" and is now available to receive applicants.
• Interested and eligible applicants and submit their applications by referencing the CFDA number(s): 47.070.
• This funding opportunity was created on Oct 19, 2018.
• Applicants must submit their applications by Jan 17, 2019. (Agency may still review applications by suitable applicants for the remaining/unused allocated funding in 2026.)
• The number of recipients for this funding is limited to 10 candidate(s).
• Eligible applicants include: Others (see text field entitled Additional Information on Eligibility for clarification).

Apply for 19 505

[Watch] Creating a grant proposal using the step-by-step wizard inside the applicant portal: