11297

Platform-Specific Optimization and Mapping of Stencil Codes through Refinement

Marcel Koster, Roland Leissa, and Sebastian Hack, Richard Membarth, Philipp Slusallek
Compiler Design Lab, Saarland University, Intel Visual Computing Institute
1st International Workshop on High-Performance Stencil Computations (HiStencils ’14), 2014

@article{koster2014platform,

   title={Platform-Specific Optimization and Mapping of Stencil Codes through Refinement},

   author={K{"o}ster, Marcel and Lei{ss}a, Roland and Hack, Sebastian and Membarth, Richard and Slusallek, Philipp},

   year={2014}

}

Download Download (PDF)   View View   Source Source   

764

views

A straightforward implementation of an algorithm in a general-purpose programming language does usually not deliver peak performance: compilers often fail to automatically tune the code for certain hardware peculiarities like memory hierarchy or vector execution units. Manually tuning the code is firstly error-prone as well as time-consuming and secondly taints the code by exposing those peculiarities to the implementation. A popular method to circumvent these problems is to implement the algorithm in a Domain-Specific Language (DSL). A DSL compiler can then automatically tune the code for the target platform. In this paper we show how to embed a DSL for stencil codes in another language. In contrast to prior approaches we only use a single language for this task. Furthermore, we offer explicit control over code refinement in the language itself which is used to specialize stencils for particular scenarios. Our first results show that our specialized programs achieve competitive performance compared to hand-tuned CUDA programs.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: