The Desktop is Dead — Long Live the Workstation?

Author(s): Aaron Knoll
Affiliation(s): SCI Institute, University of Utah


Abstract:

We explore shared-memory workstations as compelling alternatatives to desktops and small clusters, for purposes of scientific visualization. With new manycore CPU hardware on the horizon and the current popularity of large-memory “fat nodes” in HPC, SMP workstations are poised to make a comeback. These machines will augment, not replace, HPC and cloud resources, providing both remote visualization and more personalized vis labs. They will be accessible anytime, anywhere on any device, running a single operating system, capable of handling all but the absolute largest scientific data. We describe current state of the art, emerging trends, and use cases that could make the SMP workstation the dominant driver of high-end scientific visualization in the next decade.