GPU Accelerated Parallel Cholesky Factorization

Liang Wang; Yi Sheng Zhang; Bin Zhu; Chi Xu; Xiao Wei Tian; Chao Wang; Jian Hua Mo; Li Jian

doi:10.4028/www.scientific.net/AMM.148-149.1370

Paper Titles

Phases Evolution and Y Solid Solution Microstructure in the Mg-6Zn-3Y Alloy Solidified under Super-High Pressure
p.1347

Improving the Granulating and Sintering Performance by Pretreating Concentrates Using Extrusion Machine
p.1351

Study on Machining Error of Globoidal Cam Profile Resulting from Motion Error of Machine Tool in Machining
p.1356

Classification of Underwater Echo Based on Fractal Theory and Learning Vector Quantization Neural Network
p.1365

GPU Accelerated Parallel Cholesky Factorization
p.1370

Application of EFDC Model in Preventing Water Quality Deterioration through Water Circulation
p.1374

Effect of Rare Earth Salt on Ceramic Membrane Formed by Micro-Arc Oxidation
p.1378

Effect of Dynamic Baking on BH Value of Ultra Low-Carbon IF Steel
p.1383

Water Lubricated Mechanism Based on Multilevel Method
p.1388

HomeApplied Mechanics and MaterialsApplied Mechanics and Materials Vols. 148-149GPU Accelerated Parallel Cholesky Factorization

GPU Accelerated Parallel Cholesky Factorization

Abstract:

One of the fundamental problems in scientific computing is to find solutions for linear equation systems. For finite element problem, Cholesky factorization is often used to solve symmetric positive definite matrices. In this paper, Cholesky factorization is massively parallelized and three different optimization methods - highly parallel factorization, tile strategy and memory scheduling are used to accelerate Cholesky factorization effectively. A novel algorithm using OpenCL is implemented. Testing on GPU shows that performance of the algorithm increases with the dimension of matrix, reaching 785.41GFlops, about 50x times speedup. Cholesky factorization is remarkably improved with OpenCL on GPU.

You might also be interested in these eBooks

Mechanical Engineering, Materials and Energy

View Preview

Info:

Periodical:

Applied Mechanics and Materials (Volumes 148-149)

Pages:

1370-1373

DOI:

https://doi.org/10.4028/www.scientific.net/AMM.148-149.1370

Citation:

Cite this paper

Online since:

December 2011

Authors:

Liang Wang, Yi Sheng Zhang, Bin Zhu, Chi Xu, Xiao Wei Tian, Chao Wang, Jian Hua Mo, Li Jian

Keywords:

Cholesky, GPU Accelerators, Matrix Factorization, OpenCL, Parallel Computing

Export:

RIS, BibTeX

Price:

Permissions CCC:

Request Permissions

Permissions PLS:

Request Permissions

Сopyright:

Citation:

References

[1] Tomov, S., Nath, R., Ltaief, H., Dongarra, J. Dense linear algebra solvers for multicore with GPU accelerators. in: Parallel & Distributed Processing, Workshops and PhD Forum (IPDPSW), 2010 IEEE International Symposium on: 2010. 1-8.

DOI: 10.1109/ipdpsw.2010.5470941

Google Scholar

[2] Lezar, E., Davidson, D. B. GPU-based Arnoldi factorisation for accelerating finite element Eigen analysis. in: Electromagnetics in Advanced Applications, 2009. ICEAA '09: 380-383.

DOI: 10.1109/iceaa.2009.5297413

Google Scholar

[3] WU Guiming, DOU Yong, WANG Miao. A Fine Grained Parallel Algorithm for the Cholesky Decomposition. Computer Engineering & Science. 2010, 32(9).

Google Scholar

[4] Woodsend, K., Gondzio, J. Hybrid MPI/OpenMP Parallel Linear Support Vector Machine Training. J. Mach. Learn. Res. 2009, 10: 1937-(1953).

Google Scholar

[5] Govindu, G., Scrofano, R., Prasanna, V. K. A library of parameterizable floating-point cores for FPGAs and their application to scientific computing. in. US: CSREA Press, 2005. 137-145.

Google Scholar

[6] Wang Liang, Zhang Yi-sheng. Study on Transient Temperature Field Parallel Computing in Cooling Control Based on a GPU Fourier Method, IEEE Xplore(CISE. 2010. 5676712) 2010(12), 1-4.

DOI: 10.1109/cise.2010.5676712

Google Scholar