James A. Ang, PNNL chief scientist for computing in the Physical and Computational Sciences Directorate, was invited to serve as the moderator for a panel in a virtual workshop focused on federally funded research and development on software for heterogeneous computing.
The invitation-only videoconference, “Software in the Era of Extreme Heterogeneity,” featured key stakeholders from government, industry, nonprofits, and academia. They discussed the objectives and recommendations of the 2019 report, “National Strategic Computing Initiative Update: Pioneering the Future of Computing”.
“We are entering an era of extreme heterogeneity with the integration of general purpose processor cores with a collection of specialized processor or accelerators for deep learning or inferencing, and other forms of data analytics,” said Ang. “This panel on software productivity addressed some of the challenges for the development of heterogeneous applications in a performant, portable, and composable manner.”
The September 22-24 workshop also focused on challenges imposed by extreme heterogeneity of emerging and future computational platforms. They also addressed how the software community must evolve to respond to the challenges being placed on high-end computing software development and sustainment in support of American leadership in science and engineering, economic competitiveness, and national security.
Robinson Pino, conference co-chair and program manager for the Advanced Scientific Computing Research (ASCR) program office in the U.S. Department of Energy's Office of Science, invited Ang to participate. The invitation stemmed from Ang’s leadership role for the Exascale Computing Project when he was the hardware technology director and was part of the leadership team to launch the project.
Ang is PNNL’s ASCR sector lead. His responsibilities include providing PNNL with senior technical leadership on the future of computing. He established and leads the Data-Model Convergence Initiative which directly supports laboratory objectives in accelerating scientific discovery and real time control of the power grid.