Exploiting Parallelism and Scalability | Apply Online

Frequently Asked Questions (FAQs)

What is the NSF "Exploiting Parallelism and Scalability" (XPS) grant opportunity?

XPS is a National Science Foundation (NSF) research grant opportunity focused on advancing the foundations of parallel computing and scalability. It was created in response to the shift away from automatic software speedups driven by faster single-core processors, and toward a world where real performance gains depend on using multicore, many-core, cloud, and warehouse-scale systems effectively.

Why was the XPS program created?

The program was created because single-processor performance improvements have largely flattened as semiconductor scaling approaches physical limits. At the same time, computing platforms have moved toward parallel hardware (multicore/many-core processors) and large-scale distributed systems (cloud and warehouse-scale computing). XPS targets the research needed to make parallelism more usable, predictable, and scalable so that performance progress can continue.

What kinds of research does XPS aim to fund?

XPS aims to fund ambitious, foundational research that can move parallel computing into a more scalable and usable era. The emphasis is on breakthroughs that improve the predictability, scalability, and practicality of parallel computing for real users and real workloads.

What is the main technical theme of XPS?

The main theme is exploiting parallelism effectively and at scale. Rather than depending on faster chips, the program focuses on new concepts, theories, and first-principles approaches that help systems and software make better use of parallel hardware and distributed computing resources.

Does XPS expect research at a specific layer of the computing stack?

No. A core message of XPS is that major progress is unlikely to come from only one layer. The program calls for collaborative, cross-cutting efforts spanning the full stack, from applications and services down to micro-architecture.

What examples of research areas are encouraged under XPS?

Work encouraged under XPS includes: abstract models and algorithms that better capture parallel execution and scalability limits; new programming models and languages that make correct, high-performing parallel software easier to write; hardware architectures designed for parallel workloads; and supporting toolchains and system software such as compilers, operating systems, and runtime systems.

What kinds of system software and tools are explicitly in scope?

The program description explicitly mentions compilers, operating systems, and runtime systems that handle scheduling, memory management, and communication coordination, as well as broader toolchain and system software that support scalable parallel execution.

How important are programming models and languages in XPS?

They are a central area of interest. XPS highlights the need for programming models and languages that make it easier to build parallel software that is both correct and high-performing, helping reduce the difficulty and unpredictability that often comes with parallel programming.

Does XPS support hardware architecture research?

Yes. The scope includes hardware architectures designed with parallel workloads in mind, along with the co-designed software layers that help those architectures deliver scalable performance in practice.

What does XPS mean by "domain- and application-specific knowledge"?

XPS expects proposals to leverage what is unique about particular problem areas rather than relying only on one-size-fits-all approaches. In other words, it highlights the value of designs and methods that exploit domain-specific structure to achieve better scalability, efficiency, or usability.

Is raw performance the only concern in XPS?

No. While scalability and performance are core goals, XPS also highlights practical constraints that strongly shape modern computing, especially energy efficiency and communication efficiency.

Why is energy efficiency emphasized in this program?

The program notes that power and thermal limits often cap achievable performance, particularly in large data centers and dense many-core systems. As a result, energy efficiency is treated as a major constraint that parallel systems and software must address.

Why is communication efficiency emphasized in XPS?

XPS highlights that moving data between cores, accelerators, memory, and distributed machines can dominate both time and energy costs. Research that addresses these communication bottlenecks is therefore aligned with the program goals.

Does XPS consider edge and cloud computing together?

Yes. The program points to the growing importance of intelligently splitting computation between edge devices and the cloud, particularly for emerging applications where latency, bandwidth, privacy, and heterogeneous distributed hardware matter.

What kinds of real-world constraints are mentioned for edge-to-cloud computing?

The program description specifically mentions latency constraints, bandwidth limits, privacy considerations, and heterogeneous hardware distributed across networks as key factors that affect how computation should be partitioned between edge devices and cloud systems.

What is the funding opportunity number for XPS?

The funding opportunity number is 16-507.

What CFDA number is associated with this opportunity?

The opportunity is associated with CFDA 47.070.

What type of NSF grant was this administratively categorized as?

It was offered as a discretionary NSF research grant within the "Science and Technology and other Research and Development" category.

When was this opportunity posted and created?

The posting and creation dates were October 20, 2015.

What was the closing date for this opportunity?

The closing date was January 19, 2016.

What was the anticipated award ceiling?

The anticipated award ceiling was $1,000,000.

How many awards were expected?

The opportunity indicated an expectation of about 21 awards.

Who was eligible to apply?

Eligibility was listed broadly as "Others," with additional eligibility details referenced in the full solicitation text.

Where are the detailed eligibility rules described?

The opportunity summary indicates that additional eligibility details are provided in the full solicitation text.

What is the program trying to change about parallel computing for users?

XPS aims to make parallel computing more predictable, scalable, and practical for real users and real workloads, suggesting a focus on usability and reliability alongside technical performance and scaling.

What does XPS say about the end of "free" performance gains?

The program notes that for many years software often benefited from automatic speedups as single-core processors improved each generation. It states that this era has largely ended, and that continued progress now depends on effectively exploiting parallelism.