Scientific modeling and simulation software is ubiquitous at Sandia National Laboratories and is integral to providing empirical justification to critical mission decisions. Models are increasingly being expressed as workflows to simplify the many steps needed in scientific analyses but keeping these models and workflows alive for the decades-long timescales needed by Sandia remains a struggle. Additionally, the manual use and (lack of) maintenance of these models creates significant risks for duplicated work and model capability loss over time from changing personnel and computing environments. To address these issues, we are building the Engineering Common Modeling Framework (ECMF), a platform for scientific model sustainment at Sandia. ECMF enables the automatic evaluation of models over time and will ensure that models created at Sandia are discoverable and ready to be revisited, extended, and reused. In this paper, we report our current and planned capabilities as well as lessons learned from our framework development process.

We assess the reproducibility of the BugsInPy dataset less than three years after its original publication. The bug dataset provides some information about the software environment in which the code should be run, but this information can be incomplete or can decay into something uninstallable over time. We rectify as many of these problems as we can and redesign the original dataset to be more easily reusable and reproducible by future research projects. Based on our experience, we offer suggestions to authors of Python artifacts to improve their reproducibility.

The purpose of this study is to discuss the experiential lessons associated with adopting scientific workflows in the Exascale Additive Manufacturing project (ExaAM) through the lens of Perceived Characteristic of Innovation (PCI). Besides the implementation, the factors we considered critical to the adoption of the workflow are provenance, sustainable automation, implementation challenges, and integration/compatibility challenges. Through conversations and interviews among the program managers, project leads, and software engineers, we have developed critical insight and strategies to overcome the obstacles and augment the successful adoption and long-term use of these workflows in ExaAM and beyond. We hope our work will pave the way for others in the research community to develop and use workflows in their respective science domains.

Robots have to take highly responsive real-time actions, driven by complex decisions involving a pipeline of sensing, perception, planning, and reaction tasks. These tasks must be scheduled on resource-constrained devices such that the performance goals and the requirements of the application are met. This is a difficult problem that requires handling multiple scheduling dimensions, and variations in computational resource usage and availability. In practice, system designers manually tune parameters for their specific hardware and application, which results in poor generalization and increases the development burden. In this work, we highlight the emerging need for scheduling CPU resources at runtime in robot systems. We use robot navigation as a case-study to understand the key scheduling requirements for such systems. Armed with this understanding, we develop a CPU scheduling framework, Catan, that dynamically schedules compute resources across different components of an app so as to meet the specified application requirements. Through experiments with a prototype implemented on ROS, we show the impact of system scheduling on meeting the application's performance goals, and how Catan dynamically adapts to runtime variations.

Hardware specialization is becoming a key enabler of energy-efficient performance. Future systems will be increasingly heterogeneous, integrating multiple specialized and programmable accelerators, each with different memory demands. Traditionally, communication between accelerators has been inefficient, typically orchestrated through explicit DMA transfers between different address spaces. More recently, industry has proposed unified coherent memory which enables implicit data movement and more data reuse, but often these interfaces limit the coherence flexibility available to heterogeneous systems. This paper demonstrates the benefits of fine-grained coherence specialization for heterogeneous systems. We propose an architecture that enables low-complexity independent specialization of each individual coherence request in heterogeneous workloads by building upon a simple and flexible baseline coherence interface, Spandex. We then describe how to optimize individual memory requests to improve cache reuse and performance-critical memory latency in emerging heterogeneous workloads. Collectively, our techniques enable significant gains, reducing execution time by up to 61% or network traffic by up to 99% while adding minimal complexity to the Spandex protocol.