Challenge benchmarks that must be conquered to sustain the gpu revolution
University of Wisconsin, Madison
4th Workshop on Emerging Applications for Manycore Architecture, 2011
@inproceedings{eama11:challenge,
author={Emily Blem and Matthew Sinclair and Karthikeyan Sankaralingam},
title={Challenge Benchmarks that must be conquered to sustain the GPU revolution},
booktitle={"{Proceedings of the 4th Workshop on Emerging Applications for Manycore Architecture}"},
bib_dl_pdf={http://bit.ly/laizPz},
bib_pubtype={Workshop},
bib_rescat={Architecture},
year={2011}
}
The shift from GPUs to GPGPUs has brought with it many changes to the GPU architecture (e.g. more caches, more concurrent kernels, better synchronization). As GPUs press further into the general-purpose domain, architects must continue to address the performance of challenging workloads. This paper presents a set of challenge benchmarks and their key performance limitations to help direct future GPU architecture research. Our study shows GPUs must develop multiple innovative architectural techniques to efficiently execute these applications to continue making inroads into general purpose computing.
November 21, 2011 by hgpu
