This year, the Practice and Experience in Advanced Research Computing, or PEARC, Conference celebrated its 10th year in Minneapolis, July 26-30, with more than 1,050 attendees gathered around the theme “Resilient Roots + Empowered Communities.” Sponsored by the Association for Computing Machinery (ACM), the PEARC Conference Series promotes collaboration and inclusivity across the research computing and cyberinfrastructure communities. It brings together participants from academia, government and industry to share knowledge, address challenges and explore innovative solutions, with support from ACM and a network of dedicated volunteers.
As in previous years, U.S. National Science Foundation ACCESS personnel participated in leadership roles, at a Gold-level booth and presenting both posters and papers, some of which were honored with awards.
The Appverse: A Community-Driven Catalog for Open OnDemand Applications – Best Full Paper: Systems Track
Authors: Joshua Winchell, Dhruva Chakravorty, Sean Anderson, Julia Ma, Alan Chalker, Lissie Fein, Andrew Pasquale, William Horka, Marinus Pennings, Karen Tomko, Keegan Smith and Alan Sussman
Open OnDemand is an open-source web portal that simplifies access to high-performance computing (HPC) resources and can be extended through customizable applications. To reduce duplicated development efforts across institutions, the Appverse provides a community-driven catalog where users can discover, evaluate, and deploy Open OnDemand apps. The platform highlights app quality, adoption and community engagement through GitHub activity and voluntary usage reporting. Since its launch earlier this year, the Appverse has already seen international adoption with 106 apps from 20 contributing institutions.
Integrating Heterogeneous HPC Resources into National Cyberinfrastructure: A Dual-SidedWorkflow Ecosystem for “ACCESS” Resource Providers and Staff – Best Full Paper: Applications Track
Authors: Esen Gokpinar Shelton, John-Paul Navarro, Dinuka De Silva, Eric Blau, Winona Snapp-Childs
This paper presents a dual-sided software ecosystem for ACCESS designed to streamline and coordinate the onboarding of heterogeneous HPC resources. Grounded in user-centered design and informed by Cognitive Load Theory, Distributed Cognition and Service Design, the system transforms a fragmented, documentation-driven process into a unified workflow, connecting Resource Provider onboarding with staff-side review, verification and support. Developed through an iterative Double Diamond design process, the interface uses a roadmap-badge-task model to make complex requirements, dependencies, responsibilities and progress visible and actionable across stakeholder groups. By embedding workflow and governance logic directly into the user experience, the approach reduces cognitive burden, strengthens coordination across distributed actors and establishes a scalable interaction model for onboarding within complex, governance-intensive cyberinfrastructure ecosystems.
Analyzing HPC Job Wait Times under Resource Scaling Using Historical Workload Data – Best Short Paper: Systems Track
Authors: Bipin Gaikwad, Shraddha Singh, Manasvita Joshi, Nikolay Simakov, Thomas Furlani, Joseph White, Abani Patra
This work presents a data-driven framework that uses historical HPC job traces and a synthetic scheduler to evaluate how hardware upgrades, such as adding GPUs or CPU cores, would affect job wait times without altering production systems. The results show that increasing GPU resources can significantly reduce wait times, while CPU-only scaling often provides limited improvement. The authors also introduce the Weighted Wait-Time Score (WWS), a metric for measuring queue performance, and a cost-aware optimization approach to help HPC centers make more effective infrastructure investment decisions.
Congratulations to all ACCESS authors on bringing your expertise to the PEARC26 conference and helping more researchers, educators and students accelerate their work through the program’s resources!