Automatic code generation and tuning for stencil kernels on modern shared memory architectures
Department of Mathematics and Computer Science, University of Basel, Klingelbergstrasse 50, 4056 Basel, Switzerland
Computer Science – Research and Development, Volume 26, Numbers 3-4, pp. 205-210, 2011
@article{christen2011automatic,
title={Automatic code generation and tuning for stencil kernels on modern shared memory architectures},
author={Christen, M. and Schenk, O. and Burkhart, H.},
journal={Computer Science-Research and Development},
pages={1–6},
year={2011},
publisher={Springer}
}
In this paper, we present Patus, a code generation and auto-tuning framework for stencil computations targeted at multi- and manycore processors, such as multicore CPUs and graphics processing units. Patus, which stands for "Parallel Autotuned Stencils," generates a compute kernel from a specification of the stencil operation and a strategy which describes the parallelization and optimization to be applied, and leverages the autotuning methodology to optimize strategy-specific parameters for the given hardware architecture.
December 18, 2011 by hgpu
