Accelerating weather and climate simulations on heterogeneous architectures
(Presentation - presented in Xi'an, China, Oct. 2016, Invited talk at the CoDesign Meeting at HPC China 2016 )
Abstract
Stencil computations form the heart of many weather and climate code. As opposed to traditional views on these computations that only consider a single function applied to a single field, weather and climate codes exhibit complex stencil chains computing dozens of input and output fields. Code transformations, such as loop tiling and loop fusion, are of key importance for the efficient implementation of stencil computations. However, their direct application to a large code base is costly and severely impacts program maintainability. While recently introduced domain-specific languages facilitate the application of such transformations, they typically still require manual tuning or auto-tuning techniques to select the transformations that yield optimal performance. We introduce MODESTO, a model-driven stencil optimization framework, that for a stencil program suggests program transformations optimized for a given target architecture. Initially, we review and categorize data locality transformations for stencil programs and introduce a stencil algebra that allows the expression and enumeration of different stencil program implementation variants. Combining this algebra with a compile-time performance model, we show how to automatically tune stencil programs. We use our framework to model the STELLA library and optimize kernels used by the COSMO atmospheric model on multi-core and hybrid CPU-GPU architectures. Compared to naive and expert-tuned variants, the automatically tuned kernels attain a 2.0-3.1x and a 1.0-1.8x speedup respectively.
Documents
download slides:
Recorded talk (best effort)
BibTeX
@misc{hoefler-codesign16, author={Torsten Hoefler}, title={{Accelerating weather and climate simulations on heterogeneous architectures}}, year={2016}, month={Oct.}, location={Xi'an, China}, note={Invited talk at the CoDesign Meeting at HPC China 2016}, source={http://www.unixer.de/~htor/publications/}, }
