Taming Irregular Computations On Heterogeneous processors

ended

In the past decade, more and more heterogeneous processors are equipping modern supercomputers. Unfortunately, despite the progress of hardware infrastructure, the utilization of heterogeneous computing is still relatively low in practice. The utilization mainly suffers from two bottlenecks: distributed memories that bring high cost for software engineering and degraded performance, and lack of heterogeneity-oriented scalable approaches for irregular problems. The objective of the project TICOH is to address the issue of currently unsatisfactory utilization of heterogeneous computing for irregular problems such as graph and sparse matrix processing. Following a multi-level approach which bridges the domains of performance measurement, benchmark data analysis, modeling, data structure construction, algorithm design and application integration, TICOH will explore best practices that toward the best performance for irregular computations on the best hardware selection. Specifically, the main focus of the project will be to (a) identify and understand bottlenecks of current heterogeneous computing, (b) benchmark and model heterogenous processors composed of CPU, GPU and high-bandwidth memories, (c) design and evaluate new data structures and algorithms for irregular problems aiming for fully use computing and memory resources, and (d) integrate and apply the newly designed approaches for high-level applications. By empirically investigating these issues, the ultimate goal of the project is to allow a broad range of real-world applications to further benefit from heterogeneous hardware in the new era. Prof. Anne C. Elster will be hosting this action at the HPC-Lab of NTNU. Her research team is known for its experience with heterogeneous computing, and is one of the main partners in the H2020 CloudLightning project. This environment will guarantee successful bidirectional transfer of knowledge.