IBM uses a data-centric computing architecture to help the U.S. Energy Department build high-performance supercomputers under a $325 million contract between the company and the agency, InformationWeek reported Aug. 7.
David Turek, vice president of technical computing at IBM, told the publication in an interview that high-performance computing capacity is measured through exascale nodes and based on the speed of applications that run in the system.
“The movement of data creates delay,” Turek was quoted as saying.
“We are using infrastructure to deal with the problem in totality,” he added.
The contract with DOE calls for the technology firm to help implement HPCs at the Lawrence Livermore National Laboratory in California and the Oak Ridge National Laboratory in Tennessee by 2017.
The effort is part of President Barack Obama’s National Strategic Computing Initiative, which aims to produce a system that can operate at least one exaflop, the publication reports.