X-Strack: Programming Environments for Scientific Computing | DE FOA 0002460

Opportunity Information: Apply for DE FOA 0002460

X-Strack: Programming Environments for Scientific Computing (DOE FOA DE-FOA-0002460) is a U.S. Department of Energy (DOE) Office of Science opportunity, run through the Advanced Scientific Computing Research (ASCR) program, that seeks basic computer science research to improve how scientific software is built and sustained for modern high-performance computing (HPC). The core focus is on new programming environments and software-stack ideas that make it easier to create, verify, validate, optimize, maintain, and run scientific codes on systems that are increasingly both heterogeneous (many kinds of compute devices) and distributed (many networked nodes). The motivation is that scientific teams are being squeezed by growing hardware complexity, growing software complexity, and the need to port and re-port large code bases as new architectures appear, all while still delivering trustworthy science and acceptable performance.

A central theme in the FOA is heterogeneity. DOE points to the current reality that leadership-class machines commonly combine CPUs and accelerators, and highlights upcoming exascale systems such as Aurora and Frontier as examples of nodes with multiple CPUs and GPUs. The announcement also anticipates an even broader mix of accelerators in future systems, including GPUs, coarse-grained reconfigurable architectures (CGRAs), field-programmable gate arrays (FPGAs), machine-learning accelerators, and processing-in-memory technologies. In this world, programming models need to do more than just "run on a GPU"; they must support execution across diverse device types and handle data movement and data layout as first-class concerns, because performance and even feasibility often hinge on where data lives and how it moves between memories and devices.

The FOA also emphasizes distribution at scale. Typical HPC platforms consist of hundreds to thousands of nodes connected by high-performance interconnects. Many are distributed-memory systems where data must be explicitly communicated across nodes (often via MPI), even if some systems offer shared or global address-space features. Scientific applications frequently require large memory footprints and rely on scaling out to many nodes to fit problem state and run in reasonable time. As node counts rise, the cost of data movement, algorithmic scaling behavior, load balance, and communication patterns become decisive design constraints. The opportunity frames these realities as fundamental drivers for new research in programming environments and tooling that can help developers cope with complexity while still extracting performance.

A productivity and sustainability message runs through the entire announcement. DOE argues that fully realizing next-generation HPC benefits depends on a development cycle that is both high-productivity and sustainable, not just a one-off port to a single machine. That development cycle includes designing new capabilities, implementing them, porting existing functionality to new platforms, verifying correctness, optimizing performance, and integrating changes into evolving code bases. The FOA highlights that verification after platform changes or code modifications can be expensive and time-consuming, yet it is critical for scientific integrity. To address this, DOE points to modern program analysis and program synthesis methods, including approaches that leverage formal methods and machine learning, as promising paths toward more automated programming environments, along with more powerful verification, validation, and repair techniques.

Two particularly stressed research directions are automation in testing and assistance in moving between parallel programming models. On testing, the FOA calls out the need for improved productivity and effectiveness in verification, explicitly identifying automated test synthesis as a high-priority area because testing and validation often sit on the critical path when codes are modified or moved to new platforms. On portability, DOE notes that no single parallel programming model is best across the full range of platforms due to differences in availability, interoperability, and performance. As a result, it places high value on research that can help programmers transition existing applications between models such as OpenMP, OpenACC, SYCL, Kokkos, RAJA, CUDA, and other vendor-specific approaches. This is framed as "parallel-programming-model translation," aimed at reducing the cost and risk of porting and enabling DOE's diverse application ecosystem to move across hardware generations without constant rewrites.

From an administrative and eligibility standpoint, this is a discretionary research funding opportunity using a cooperative agreement mechanism under CFDA 81.049 (Office of Science). A key process requirement is that a pre-application must be submitted; applications without a pre-application are rejected without review. Eligibility is broad in principle, but with several important carve-outs. Nonprofits classified as 501(c)(4) that engaged in lobbying after December 31, 1995 are not eligible. The FOA also warns that individual applicants are unlikely to have sufficient resources for the work and should explicitly address how they will meet the objectives if applying. Non-domestic applicants are permitted, but they are advised that successful proposals typically need to justify why the applicant offers unique skills or resources not available among domestic organizations.

The FOA draws a firm line against commercial activity. It does not support commercial work, and proposals from for-profit organizations that are tied to current business activities or uses are considered commercial and will be declined. It also excludes work that is or has been supported by, or proposed to, SBIR/STTR programs, treating that overlap as commercial activity and declining it without merit review. For-profit applicants that are otherwise eligible must include a short (up to 200 words) appendix statement explaining how the project advances scientific understanding at a basic and fundamental level, reinforcing that the program is meant to fund foundational research rather than product development.

Special rules apply to federally affiliated entities. DOE/NNSA National Laboratories can apply as lead organizations or as team members, but they cannot be included as subrecipients under another organization's application; if selected as a lead, funding is handled through DOE's Field-Work Proposal system with contracting officer authorization steps described in the FOA. Non-DOE/NNSA FFRDCs cannot apply as prime recipients but may participate as subrecipients; if selected, DOE may route support via an inter-agency award to the FFRDC's sponsoring agency rather than through a standard subaward. Other federal agencies are not eligible either as applicants or as subrecipients.

Key opportunity metadata included in the notice: the Funding Opportunity Number is DE-FOA-0002460, the title is "X-Strack: Programming Environments for Scientific Computing," the sponsoring agency is the DOE Office of Science, and the original closing date listed is April 12, 2021. The listed award ceiling is $300,000, and the funding instrument is a cooperative agreement. Overall, the opportunity is positioned as a focused investment in foundational computer science research that can materially improve the way scientific software is developed, verified, and ported across the heterogeneous, distributed HPC systems that DOE science increasingly depends on.

• The Office of Science in the science and technology and other research and development sector is offering a public funding opportunity titled "X-Strack: Programming Environments for Scientific Computing" and is now available to receive applicants.
• Interested and eligible applicants and submit their applications by referencing the CFDA number(s): 81.049.
• This funding opportunity was created on 2021-01-28.
• Applicants must submit their applications by 2021-04-12. (Agency may still review applications by suitable applicants for the remaining/unused allocated funding in 2026.)
• Each selected applicant is eligible to receive up to $300,000.00 in funding.
• Eligible applicants include: Others.

Apply for DE FOA 0002460

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