6482

Communication-avoiding QR decomposition for GPUs

Michael Anderson, Grey Ballard, James Demmel, Kurt Keutzer
UC Berkeley: Department of Electrical Engineering and Computer Science, Berkeley, CA USA
IEEE International Parallel & Distributed Processing Symposium (IPDPS), 2011

@inproceedings{anderson2011communication,

   title={Communication-avoiding QR decomposition for GPUs},

   author={Anderson, M. and Ballard, G. and Demmel, J. and Keutzer, K.},

   booktitle={Parallel & Distributed Processing Symposium (IPDPS), 2011 IEEE International},

   pages={48–58},

   year={2011},

   organization={IEEE}

}

Download Download (PDF)   View View   Source Source   

716

views

We describe an implementation of the Communication-Avoiding QR (CAQR) factorization that runs entirely on a single graphics processor (GPU). We show that the reduction in memory traffic provided by CAQR allows us to outperform existing parallel GPU implementations of QR for a large class of tall-skinny matrices. Other GPU implementations of QR handle panel factorizations by either sending the work to a general-purpose processor or using entirely bandwidth-bound operations, incurring data transfer overheads. In contrast, our QR is done entirely on the GPU using compute-bound kernels, meaning performance is good regardless of the width of the matrix. As a result, we outperform CULA, a parallel linear algebra library for GPUs by up to 17x for tall-skinny matrices and Intel’s Math Kernel Library (MKL) by up to 12x. We also discuss stationary video background subtraction as a motivating application. We apply a recent statistical approach, which requires many iterations of computing the singular value decomposition of a tall-skinny matrix. Using CAQR as a first step to getting the singular value decomposition, we are able to get the answer 3x faster than if we use a traditional bandwidth-bound GPU QR factorization tuned specifically for that matrix size, and 30x faster than if we use Intel’s Math Kernel Library (MKL) singular value decomposition routine on a multicore CPU.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: