<span class="var-sub_title">Applying the Execution-Cache-Memory Model: Current State of Practice</span> SC18 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Applying the Execution-Cache-Memory Model: Current State of Practice


Authors: Georg Hager (University of Erlangen-Nuremberg, Erlangen Regional Computing Center), Jan Eitzinger (University of Erlangen-Nuremberg, Erlangen Regional Computing Center), Julian Hornich (University of Erlangen-Nuremberg, Erlangen Regional Computing Center), Francesco Cremonesi (Swiss Federal Institute of Technology in Lausanne), Christie L. Alappat (University of Erlangen-Nuremberg, Erlangen Regional Computing Center), Thoams Roehl (University of Erlangen-Nuremberg), Gerhard Wellein (University of Erlangen-Nuremberg, Erlangen Regional Computing Center)

Abstract: The ECM (Execution-Cache-Memory) model is an analytic, resource-based performance model for steady-state loop code running on multicore processors. Starting from a machine model, which describes the interaction between the code and the hardware, and static code analysis, it allows an accurate prediction of the runtime of sequential loop code. Together with a scaling assumption, it also gives a performance scaling prediction. This poster summarizes the current state of practice in constructing and applying the ECM model, points out problems and open questions, and applies the model to three new and nontrivial use cases. For the first time, overlap assumptions for all relevant CPU architectures in high performance computing are presented.

Best Poster Finalist (BP): yes

Poster: pdf
Poster summary: PDF
Reproducibility Description Appendix: PDF


Back to Poster Archive Listing