Reproducible stencil compiler benchmarks using prova

The stencil pattern represents a vast variety of applications, ranging from geophysics to medical science. In application codes, the stencil kernel is often the part where most of the time is spent, thus forcing an efficient parallel implementation of it. On the other side, we know that stencil comp...

Full description

Saved in:
Bibliographic Details
Published in:Future generation computer systems 2019-03, Vol.92, p.933-946
Main Authors: Guerrera, Danilo, Maffia, Antonio, Burkhart, Helmar
Format: Article
Language:English
Subjects:
HPC
Performance engineering
Reproducibility
Reproducible research
Roofline
Stencil
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The stencil pattern represents a vast variety of applications, ranging from geophysics to medical science. In application codes, the stencil kernel is often the part where most of the time is spent, thus forcing an efficient parallel implementation of it. On the other side, we know that stencil computations are often memory-bound, which requires sophisticated parallelization techniques to get scalable solutions. In this paper, we present the results of stencil benchmark experiments run on different systems using the prova! tool we are currently implementing. prova! aims for reproducible performance experiments and makes collaborative stencil benchmarking feasible through web repositories and interfaces. •This paper shows how to repeat and replicate stencil experiments on different systems.•Effort to implement a collaborative ecosystem for reproducible stencil benchmarking.•Performance studies and reproducibility issues can be combined.•PROVA! can help you since it hides all the tedious task related to make your experiments reproducible.
ISSN:0167-739X
1872-7115
DOI:10.1016/j.future.2018.05.023