The 2025 Global Open OnDemand (GOOD) Conference was held in March and featured a range of contributions from the U.S. National Science Foundation (NSF) ACCESS community – highlighting advancements in dashboard development and system integration. Researchers and students presented their latest work – demonstrating the power of collaboration in the high-performance computing (HPC) community.
Purdue REU Students Enhance Anvil OnDemand Dashboard
Richie Tan, a computer science student at Purdue University and participant of the Rosen Center for Advanced Computing’s (RCAC) 2024 Anvil Research Experience for Undergraduates (REU) Summer program, presented his work at the conference. Tan showcased improvements that his team – including Anjali Rajesh, also an REU participant – made to the Anvil OnDemand dashboard, which is now in production.
“With the support of everyone at the Rosen Center for Advanced Computing, we were able to deploy an enhanced version of the Anvil Open OnDemand dashboard earlier this year,” Tan said. “Participating in the NSF-funded Anvil REU program last summer helped introduce me to high performance computing and provided an engaging experience where I could apply my skills and knowledge in a real-world environment to build something useful for the community. I am looking forward to continuing iterating and adding features to our dashboard and to release the next version of the Anvil Open OnDemand dashboard soon.”
Tan’s GOOD presentation is available online.
Texas A&M University Explores Open OnDemand Integrations
Texas A&M University (TAMU) research was also featured at the conference with two presentations on their advancements in Open OnDemand implementations within their HPC systems. These talks explored innovative approaches to improving user accessibility and resource management across TAMU’s computing infrastructure.
Michael Dickens, a research scientist at TAMU High Performance Research Computing, presented a talk entitled Using a Passenger App for Displaying GPU Node Configuration and Resource Availability. His talk focused on how real-time status updates on the configuration and availability of HPC resources can help users make informed decisions when selecting parameters for interactive apps.
“GPU nodes are sometimes in short supply – with limited or no GPUs immediately available,” Dickens said. “By displaying current information—such as GPU type and quantity, available CPU cores and memory—users can better match their needs to available resources – reducing wait times in job queues and improving overall efficiency.”
Another TAMU talk at the GOOD conference was presented by Marinus Pennings, associate director of user support and helpdesk at TAMU High Performance Research Computing. Pennings’ Drona Workflow Engine presentation discussed how the TAMU team’s new experimental customizable dashboard allows users to manage their resources and connect directly with a help desk. His talk also introduced the Drona workflow engine— a framework designed for creating, managing, sharing, and executing tailored workflows.
“Drona simplifies the HPC experience by handling much of the underlying complexity, allowing researchers to concentrate on their scientific work,” Pennings explained. “It supports custom environment setup and automatic resource selection; with a few keystrokes, researchers can develop and run AI workloads across different types of accelerators.”
A Look Ahead
“As Open OnDemand continues to evolve, contributions from NSF ACCESS community members are helping to simplify HPC resource access,” said Ohio Supercomputer Center (OSC) Director of Strategic Programs Alan Chalker, who helped organize the conference. “Likewise, as part of the ACCESS program, the Open OnDemand team will continue to ensure that ACCESS resources with Open OnDemand benefit from advances by the Open OnDemand community towards making HPC more user-friendly than ever before.”
For more information about the GOOD conference, visit the OSC website.