This tutorial introduces the Cache-Aware Roofline Model and discusses its principles when modeling the performance of Intel® CPU and GPU devices. It showcases how Cache-Aware Roofline Model implementation in Intel® Advisor can be used to drive application optimization. The demonstration uses epistasis detection as a case study, which is an important application in bioinformatics. For lntel® GPUs, see how Cache-Aware Roofline Model can be used to detect execution bottlenecks and provide useful hints on which type of optimizations to apply to fully exploit the device capabilities. The guidelines provided by this model were fundamental to achieve speedups of more than 20x when compared to the baseline code.

Speaker

Aleksandar Ilic is an assistant professor at Instituto Superior Técnico (IST), Universidade de Lisboa, and a senior researcher of the INESC-ID, Portugal. He contributed to more than 50 scientific publications. His research interests include high-performance and energy-efficient computing, and modeling of heterogeneous systems.

Rafael Campos is a young researcher at Instituto de Engenharia de Sistemas e Computadores R&D (INESC-ID), as part of the HPCAS group. His main interests are performance modeling of heterogeneous systems, with a focus on performance optimization of bioinformatics applications, and roofline modeling of high-performance heterogeneous CPU and GPU systems.

Diogo Marques is a member of the HPCAS group at Instituto de Engenharia de Sistemas e Computadores R&D (INESC-ID). His research interests include the modeling of multicore and heterogeneous systems. His work contributed to improving the accuracy of the Cache-Aware Roofline Model by proposing the memory metrics and scaled roofs presented in Intel Advisor.