Parallel programming with inductive synthesis
University of California, Berkeley
3rd USENIX conference on Hot topic in parallelism (HotPar’11), 2011
@inproceedings{barman2011parallel,
title={Parallel programming with inductive synthesis},
author={Barman, S. and Bodik, R. and Jain, S. and Pu, Y. and Srivastava, S. and Tung, N.},
booktitle={Proceedings of the 3rd USENIX conference on Hot topic in parallelism},
pages={14–14},
year={2011},
organization={USENIX Association}
}
We show that program synthesis can generate GPU algorithms as well as their optimized implementations. Using the scan kernel as a case study, we describe our evolving synthesis techniques. Relying on our synthesizer, we can parallelize a serial problem by transforming it into a scan operation, synthesize a SIMD scan algorithm, and optimize it to reduce memory conflicts.
December 19, 2011 by hgpu
