A Taxonomy of HPDA Algorithms
Workshop: IA^3 2017 - 7th Workshop on Irregular Applications: Architectures and Algorithms
Abstract: This talk will be the first public presentation of the rationale, methodology and key findings of a late-2016, government-funded study Hyperion Research (then called IDC) conducted, entitled Developing a Real-World Taxonomy of the Underlying Mathematical, Algorithmic, and Technological Aspects of HPDA Applications. For this study, Hyperion first consulted public- and private-sector experts to develop a taxonomy matrix aimed at matching advanced analytics (HPDA) application types with users’ preferred algorithms. Next came in-depth interviews with scientific, academic and industrial users to identify the hardware-software requirements of their applications and the attributes of the algorithms that generate those requirements. The final step was a report designed as a reference tool for HPDA users (including non-HPC specialists) from the broad spectrum of application domains investigated in the study.