After more than a decade of discussion, planning, designing and system building, 2021 is expected to be the year that exascale-class computing finally arrives in the U.S. Hewlett Packard Enterprise is at the epicenter of this effort. The company is the prime contractor or system architect for the first three exascale systems to be delivered to U.S. Department of Energy national labs, starting with the Frontier supercomputer at Oak Ridge National Laboratory, scheduled for installation by the end of this year.
In this interview, HPE Senior HPC & AI Product Manager Neil Dey provides an overview and insights into the challenges, complexities and key components of an exascale system – from the need for a flexible infrastructure able to support a variety of processors, to the demands placed on the interconnect, to cooling, power consumption and software.
Dey also discusses an intended consequence of HPE’s exascale work, which is the adaptation of exascale-related techniques and technologies for the broader HPC and advanced computing markets.
