Skip to main content

The 33rd annual IPDPS took place in Rio de Janerio May 20 to 24

A team of researchers from the National Center for Computational Sciences (NCCS) at the US Department of Energy’s (DOE) Oak Ridge National Laboratory (ORNL) received the “Best Paper” award in the Advances in Parallel and Distributed Computational Models (APDCM) workshop of the 33rd IEEE International Parallel and Distributed Processing Symposium (IPDPS), an annual forum that brings together researchers from around the world to present topics relating to parallel computation. This year’s IPDPS took place May 20–24 in Rio de Janeiro, Brazil.

The paper, titled “AHEAD: Tool for Projecting Next-Generation Hardware Enhancements on GPU-AccelerateSystems,” was authored by Christopher Zimmer and Sudharshan S. Vazhkudai from the NCCS, as well as Hazem A. Abdelhafez and Matei Ripeanu from the University of British Columbia. Abdelhafez, who interned for Zimmer and Vazhkudai at ORNL last summer, presented the paper. “AHEAD” presents a node modeling tool to ask “what-if” questions, particularly focused on the intra-node connectivity between CPUs and GPUs. Researchers presented “AHEAD” on Monday, May 20 at the APDCM.

ORNL researchers won the best paper award at the 21st APDCM Workshop of the IPDPS annual meeting

Later in the week, Vazhkudai presented “Data Jockey: Automatic Data Management for HPC Multi-Tiered Storage Systems,” a paper co-authored by Woong Shin, Christopher D. Brumgard, Bing Xie, and Sarp Oral of ORNL, as well as Lavanya Ramakrishnan and Devarshi Ghoshal of Lawrence Berkeley National Laboratory. The paper details “Data Jockey,” a data management system for HPC multi-tiered storage systems that more efficiently automates the movement and placement of bulk data. In addition to his work with Data Jockey, Oral participated in a study that was discussed in the File Systems Session, along with Feiyi Wang.

Other ORNL researchers participating in the IPDPS meeting included Jeffrey Vetter, whose co-authored study appeared in the Accelerating Neural Networks Session, and Edmond Chow and Jack Dongarra, in the GPU Computing I Session.

A full list of speakers, workshops, and posters can be found at https://www.ipdps.org/.

UT Battelle LLC manages ORNL for DOE’s Office of Science. The Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, please visit https://energy.gov/science.