Argonne Researchers Give Presentations at Data Management Conference

Talks on data management, artificial intelligence and more at the International Conference on Scalable Scientific Data Management (SSDBM).

Aug. 29, 2025 — Researchers from the U.S. Department of Energy’s Argonne National Laboratory participated in the 37th International Conference on Scalable Scientific Data Management (SSDBM) on June 23–25, 2025, in Columbus, Ohio. SSDBM is an annual event that brings together experts to share advances in scientific data management.

Robert Ross, senior computer scientist and deputy division director in Argonne’s Mathematics and Computer Science (MCS) division, gave a keynote address titled ​“Supporting Science Through Data Management Software.” He highlighted how AI, increasingly complex workflows, and geographically distributed resources are driving the need for new data management tools in computational science.

MCS division computer scientists Bogdan Nicolae and Franck Cappello, alongside colleagues from the Illinois Institute of Technology, presented a paper titled ​“DTIO: Data Stack for AI-driven Workflows.” DTIO is a scalable I/O runtime that integrates high-performance computing, big data and machine learning workflows through intelligent interface mapping.

Nicolae and colleagues from the Illinois Institute of Technology also presented a poster titled ​“Accelerate LLM Inference with Asynchronous Model Load.” The researchers described their new method that improves hardware efficiency by prefetching model and overlapping computation with data movement.

MCS division postdoctoral appointee Amal Gueroudji co-authored a poster titled ​“Scalable GPU Performance Variability Analysis framework” with partners from Texas State University, the University of Wisconsin–Madison and DOE’s Sandia National Laboratory. The new framework uses distributed data analysis to reduce memory use, avoid central bottlenecks and enable low-latency exploration of high-dimensional trace data.

Four researchers from the Argonne Leadership Computing Facility (ALCF) — Eric Pershey, Ben Leonard, Brian Toonen and Peter Upton — and a researcher from DePaul University presented a paper titled ​“A Big Data Approach for Efficient Processing of Machine Operational Data.” Their new tool, called ODPS, helps in debugging computing clusters.

Venkat Vishwanath, the AI lead at the ALCF, served as program co-chair of the Full Papers Program Committee. The ALCF is a DOE Office of Science user facility.

The proceedings of the conference are available online here.

About ALCF

The Argonne Leadership Computing Facility provides supercomputing capabilities to the scientific and engineering community to advance fundamental discovery and understanding in a broad range of disciplines. Supported by the U.S. Department of Energy’s (DOE’s) Office of Science, Advanced Scientific Computing Research (ASCR) program, the ALCF is one of two DOE Leadership Computing Facilities in the nation dedicated to open science.

Source: ALCF