9129

Optimizing Sparse Matrix-Matrix Multiplication for the GPU

Steven Dalton, Nathan Bell, Luke N. Olson
Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL 61801
University of Illinois at Urbana-Champaign, Technical Report, 2013
@article{dalton2013optimizing,

   title={Optimizing Sparse Matrix-Matrix Multiplication for the GPU},

   author={Dalton, Steven and Bell, Nathan and Olson, Luke N},

   journal={Matrix},

   volume={3},

   year={2013}

}

Download Download (PDF)   View View   Source Source   

743

views

Sparse matrix-matrix multiplication (SpMM) is a key operation in numerous areas from information to the physical sciences. Implementing SpMM efficiently on throughput-oriented processors, such as the graphics processing unit (GPU), requires the programmer to expose substantial fine-grained parallelism while conserving the limited off-chip memory bandwidth. Balancing these concerns, we decompose the SpMM operation into three, highly-parallel phases: expansion, sorting, and compression, and introduce a set of complementary bandwidth-saving performance optimizations. Our implementation is fully general and our optimizations lead to substantial efficiencies for a SpMM product.
VN:F [1.9.22_1171]
Rating: 0.0/5 (0 votes cast)

* * *

* * *

Follow us on Twitter

HGPU group

1889 peoples are following HGPU @twitter

Like us on Facebook

HGPU group

418 people like HGPU on Facebook

HGPU group © 2010-2016 hgpu.org

All rights belong to the respective authors

Contact us: