LSU’s STE||AR Group Aligns High-Performance Computing Stars for SC20 Conference

The LSU STE||AR Group, which develops high-performance computing solutions and is housed in the Center for Computation & Technology (CCT), has been invited to host a panel about their work and the future of the field at the world’s premiere supercomputing conference, SC20. For the first time, the eleven-day conference will take place virtually, due to COVID-19, this November.

  

 

BATON ROUGE, September 1, 2020 — “We are excited and proud to have the trust of the community to organize a panel about task-based algorithms and applications that will bring together experts within the field of high-performance computing to present their ideas about the future of computation,” Hartmut Kaiser, leader of the STE||AR Group, said. “This trust demonstrates that the community recognizes our work and our results in the field.”
 
The LSU-led panel will be held on Wednesday, November 18 and address new challenges posed by exascale (read, extremely fast; faster than the most powerful supercomputers used today) system architectures that result in difficulties in using traditional distributed-memory runtimes. Task-based programming models show promise in overcoming these challenges, including the inherent load-balancing issues.

 

The LSU STE||AR Group is the lead developer of HPX, short for High Performance ParalleX, in the world.

The panel will be moderated by Patrick Diehl, LSU STE||AR Group researcher and an expert on asynchronous many-task (parallel) systems. The invited panelists will explore the advantages of task-based programming models on modern and future high-performance computing systems from an industry, university, and national lab perspective and discuss several concrete examples. Among the panelists are Laxmikant Kale, director of the Parallel Programming Laboratory at the University of Illinois at Urbana-Champaign and one of the lead developers of Charm++, which is based on C++-, both object-oriented programming languages; Irina Demeshko, a design team leader at the Los Alamos National Laboratory; Bryce Adelstein-Lelbach, one of the initial developers of the HPX parallel runtime system and a core contributor to C++, now at NVIDIA; Keno Fischer, a key developer of the Julia programming language and co-founder of Julia Computing; and Alice Koniges, a computer scientist at the University of Hawaii and associate editor of the International Journal of High Performance Computing Applications.
 
Joining them in November will be LSU computer engineering alumna Zahra Khatami who designs and develops distributed database systems for Oracle. After graduating from LSU with her Ph.D. in 2017, Khatami has been collaborating with the STE||AR Group on several HPX projects, one of which leveraged artificial intelligence to handle runtime parameters more efficiently. The LSU STE||AR Group is the lead developer of HPX, short for High Performance ParalleX, in the world.
 
“When I joined Oracle, I realized scalability plays an important role when dealing with very large numbers of tasks,” Khatami said. “I then appreciated the power of the asynchronous many-task-based runtimes I’d learned about during my Ph.D. Attending as a panelist this November will give me a chance to interact with experts in this field in both academia and industry and discover more about how we can resolve practical scalability and compatibility problems.”
 
LSU’s senior scientist Hartmut Kaiser, finally, is probably best known through his involvement in open-source software projects and as the author of several C++ libraries used by thousands of developers worldwide. He is also a voting member of the ISO C++ standards committee.

 

For more details on the panel, click here.

 

 

Elsa Hahne
LSU Office of Research & Economic Development
ehahne@lsu.edu