Distributed computing in 2030 will be defined by the need to process vast swaths of data for insights in a timely manner. Minimizing data movement to curtail energy consumption in an energy-conscious Earth will be the overriding constraint. The compute infrastructure will be a seamless hierarchy of compute centers from edge to geo-distributed mega-datacenters. Each compute center will contain a large number of heterogeneous hardware accelerators, and tasks of unprecedented small granularity will seamlessly ship computation to where data is. To further minimize data movement, key data will be replicated or re-materialized on demand. The computational environment will be highly dynamic, with the constant introduction of new classes of accelerators for barely-emerging workloads, and of new applications/protocols that could benefit from yet-to-be-conceived accelerators.
To attain this reality, the ACE Center is organized into five themes:
(Courtesy of Zhengya Zhang)
(Courtesy of Steven Swanson)
(Courtesy of Manya Ghobadi)
(Courtesy of Edward Suh)