Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method
Abstract
In this work, we present a parallel computing strategy for a hybridizable discontinuous Galerkin (HDG) nested geometric multigrid (GMG) solver. Parallel GMG solvers require a combination of coarse-grain and fine-grain parallelism to improve time-to-solution performance. In this work we focus on fine-grain parallelism. We use Intel's second generation Xeon Phi (Knights Landing) manycore processor. The GMG method achieves ideal convergence rates of 0.2 or less, for high polynomial orders. A matrix free (assembly free) technique is exploited to save considerable memory usage and increase arithmetic intensity. HDG enables static condensation, and due to the discontinuous nature of the discretization, we developed a matrix vector multiply routine that does not require any costly synchronizations or barriers. Our algorithm is able to attain 80% of peak bandwidth performance for higher order polynomials. This is possible due to the data locality inherent in the HDG method. Very high performance is realized for high order schemes, due to good arithmetic intensity, which declines as the order is reduced.
- Authors:
-
- Rice Univ., Houston, TX (United States). Dept. of Computational and Applied Mathematics
- Univ. of Buffalo, Buffalo, NY (United States). Dept. of Computer Science and Engineering
- Argonne National Lab. (ANL), Portland, OR (United States). Computer Science and Mathematics Division
- Publication Date:
- Research Org.:
- Argonne National Lab. (ANL), Argonne, IL (United States)
- Sponsoring Org.:
- National Science Foundation (NSF)
- OSTI Identifier:
- 1542582
- Grant/Contract Number:
- AC02-06CH11357
- Resource Type:
- Accepted Manuscript
- Journal Name:
- SIAM Journal on Scientific Computing
- Additional Journal Information:
- Journal Volume: 41; Journal Issue: 2; Journal ID: ISSN 1064-8275
- Publisher:
- SIAM
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; discontinuous Galerkin methods; high performance computing; multigrid; accelerators
Citation Formats
Fabien, Maurice S., Knepley, Matthew G., Mills, Richard T., and Rivière, Béatrice M. Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method. United States: N. p., 2019.
Web. doi:10.1137/17M1128903.
Fabien, Maurice S., Knepley, Matthew G., Mills, Richard T., & Rivière, Béatrice M. Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method. United States. https://doi.org/10.1137/17M1128903
Fabien, Maurice S., Knepley, Matthew G., Mills, Richard T., and Rivière, Béatrice M. Tue .
"Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method". United States. https://doi.org/10.1137/17M1128903. https://www.osti.gov/servlets/purl/1542582.
@article{osti_1542582,
title = {Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method},
author = {Fabien, Maurice S. and Knepley, Matthew G. and Mills, Richard T. and Rivière, Béatrice M.},
abstractNote = {In this work, we present a parallel computing strategy for a hybridizable discontinuous Galerkin (HDG) nested geometric multigrid (GMG) solver. Parallel GMG solvers require a combination of coarse-grain and fine-grain parallelism to improve time-to-solution performance. In this work we focus on fine-grain parallelism. We use Intel's second generation Xeon Phi (Knights Landing) manycore processor. The GMG method achieves ideal convergence rates of 0.2 or less, for high polynomial orders. A matrix free (assembly free) technique is exploited to save considerable memory usage and increase arithmetic intensity. HDG enables static condensation, and due to the discontinuous nature of the discretization, we developed a matrix vector multiply routine that does not require any costly synchronizations or barriers. Our algorithm is able to attain 80% of peak bandwidth performance for higher order polynomials. This is possible due to the data locality inherent in the HDG method. Very high performance is realized for high order schemes, due to good arithmetic intensity, which declines as the order is reduced.},
doi = {10.1137/17M1128903},
journal = {SIAM Journal on Scientific Computing},
number = 2,
volume = 41,
place = {United States},
year = {Tue Mar 12 00:00:00 EDT 2019},
month = {Tue Mar 12 00:00:00 EDT 2019}
}
Web of Science
Works referenced in this record:
Parallel multigrid smoothing: polynomial versus Gauss–Seidel
journal, July 2003
- Adams, Mark; Brezina, Marian; Hu, Jonathan
- Journal of Computational Physics, Vol. 188, Issue 2
Multigrid Algorithms for $hp$-Discontinuous Galerkin Discretizations of Elliptic Problems
journal, January 2015
- Antonietti, Paola F.; Sarti, Marco; Verani, Marco
- SIAM Journal on Numerical Analysis, Vol. 53, Issue 1
Block-asynchronous Multigrid Smoothers for GPU-accelerated Systems
journal, January 2012
- Anzt, Hartwig; Tomov, Stanimire; Gates, Mark
- Procedia Computer Science, Vol. 9
Unified Analysis of Discontinuous Galerkin Methods for Elliptic Problems
journal, January 2002
- Arnold, Douglas N.; Brezzi, Franco; Cockburn, Bernardo
- SIAM Journal on Numerical Analysis, Vol. 39, Issue 5
Load Balancing for Adaptive Multigrid Methods
journal, July 1998
- Bastian, Peter
- SIAM Journal on Scientific Computing, Vol. 19, Issue 4
Additive and multiplicative multi-grid — A comparison
journal, December 1998
- Bastian, P.; Wittum, G.; Hackbusch, W.
- Computing, Vol. 60, Issue 4
Frequency domain behavior of a set of parallel multigrid smoothing operators
journal, January 1990
- Benson, Maurice W.
- International Journal of Computer Mathematics, Vol. 36, Issue 1-2
Multi-level adaptive solutions to boundary-value problems
journal, May 1977
- Brandt, Achi
- Mathematics of Computation, Vol. 31, Issue 138
Multigrid methods for the symmetric interior penalty method on graded meshes
journal, June 2009
- Brenner, S. C.; Cui, J.; Sung, L. -Y.
- Numerical Linear Algebra with Applications, Vol. 16, Issue 6
Convergence of Multigrid Algorithms for Interior Penalty Methods
journal, May 2005
- Brenner, Susanne C.; Zhao, Jie
- Applied Numerical Analysis & Computational Mathematics, Vol. 2, Issue 1
Sparse approximate inverse smoothers for geometric and algebraic multigrid
journal, April 2002
- Bröker, O.
- Applied Numerical Mathematics, Vol. 41, Issue 1
From h to p efficiently: Strategy selection for operator evaluation on hexahedral and tetrahedral elements
journal, April 2011
- Cantwell, C. D.; Sherwin, S. J.; Kirby, R. M.
- Computers & Fluids, Vol. 43, Issue 1
A superconvergent LDG-hybridizable Galerkin method for second-order elliptic problems
journal, May 2008
- Cockburn, Bernardo; Dong, Bo; Guzmán, Johnny
- Mathematics of Computation, Vol. 77, Issue 264
A Hybridizable Discontinuous Galerkin Method for Steady-State Convection-Diffusion-Reaction Problems
journal, January 2009
- Cockburn, Bernardo; Dong, Bo; Guzmán, Johnny
- SIAM Journal on Scientific Computing, Vol. 31, Issue 5
Multigrid for an HDG method
journal, October 2013
- Cockburn, B.; Dubois, O.; Gopalakrishnan, J.
- IMA Journal of Numerical Analysis, Vol. 34, Issue 4
Unified Hybridization of Discontinuous Galerkin, Mixed, and Continuous Galerkin Methods for Second Order Elliptic Problems
journal, January 2009
- Cockburn, Bernardo; Gopalakrishnan, Jayadeep; Lazarov, Raytcho
- SIAM Journal on Numerical Analysis, Vol. 47, Issue 2
The speed of convergence of one iterative process
journal, January 1964
- Fedorenko, R. P.
- USSR Computational Mathematics and Mathematical Physics, Vol. 4, Issue 3
A multilevel discontinuous Galerkin method
journal, September 2003
- Gopalakrishnan, J.; Kanschat, G.
- Numerische Mathematik, Vol. 95, Issue 3
A convergent multigrid cycle for the hybridized mixed method
journal, September 2009
- Gopalakrishnan, Jayadeep; Tan, Shuguang
- Numerical Linear Algebra with Applications, Vol. 16, Issue 9
Two-Level Fourier Analysis of a Multigrid Approach for Discontinuous Galerkin Discretization
journal, January 2003
- Hemker, P. W.; Hoffmann, W.; van Raalte, M. H.
- SIAM Journal on Scientific Computing, Vol. 25, Issue 3
Fourier two-level analysis for discontinuous Galerkin discretization with linear elements
journal, June 2004
- Hemker, P. W.; Hoffmann, W.; van Raalte, M. H.
- Numerical Linear Algebra with Applications, Vol. 11, Issue 56
Fourier two-level analysis for higher dimensional discontinuous Galerkin discretisation
journal, October 2004
- Hemker, P. W.; van Raalte, M. H.
- Computing and Visualization in Science, Vol. 7, Issue 3-4
Efficiency of high-order elements for continuous and discontinuous Galerkin methods: EFFICIENCY OF HIGH-ORDER ELEMENTS FOR CG AND DG
journal, July 2013
- Huerta, Antonio; Angeloski, Aleksandar; Roca, Xevi
- International Journal for Numerical Methods in Engineering, Vol. 96, Issue 9
p-Multigrid solution of high-order discontinuous Galerkin discretizations of the compressible Navier–Stokes equations
journal, July 2005
- Fidkowski, Krzysztof J.; Oliver, Todd A.; Lu, James
- Journal of Computational Physics, Vol. 207, Issue 1
Exploiting Batch Processing on Streaming Architectures to Solve 2D Elliptic Finite Element Problems: A Hybridized Discontinuous Galerkin (HDG) Case Study
journal, November 2013
- King, James; Yakovlev, Sergey; Fu, Zhisong
- Journal of Scientific Computing, Vol. 60, Issue 2
To CG or to HDG: A Comparative Study
journal, July 2011
- Kirby, Robert M.; Sherwin, Spencer J.; Cockburn, Bernardo
- Journal of Scientific Computing, Vol. 51, Issue 1
A generic interface for parallel cell-based finite element operator application
journal, June 2012
- Kronbichler, Martin; Kormann, Katharina
- Computers & Fluids, Vol. 63
On Hybrid Multigrid-Schwarz Algorithms
journal, December 2007
- Loisel, Sébastien; Nabben, Reinhard; Szyld, Daniel B.
- Journal of Scientific Computing, Vol. 36, Issue 2
Hybrid Multigrid/Schwarz Algorithms for the Spectral Element Method
journal, July 2005
- Lottes, James W.; Fischer, Paul F.
- Journal of Scientific Computing, Vol. 24, Issue 1
The hp -multigrid method applied to hp -adaptive refinement of triangular grids: THE hp -MULTIGRID METHOD APPLIED TO hp -ADAPTIVE REFINEMENT
journal, March 2010
- Mitchell, William F.
- Numerical Linear Algebra with Applications, Vol. 17, Issue 2-3
A Hybridized Discontinuous Galerkin Method for the Nonlinear Korteweg–de Vries Equation
journal, November 2015
- Samii, Ali; Panda, Nishant; Michoski, Craig
- Journal of Scientific Computing, Vol. 68, Issue 1
Solving unsymmetric sparse systems of linear equations with PARDISO
journal, April 2004
- Schenk, Olaf; Gärtner, Klaus
- Future Generation Computer Systems, Vol. 20, Issue 3
From h to p efficiently: Implementing finite and spectral/hp element methods to achieve optimal performance for low- and high-order discretisations
journal, July 2010
- Vos, Peter E. J.; Sherwin, Spencer J.; Kirby, Robert M.
- Journal of Computational Physics, Vol. 229, Issue 13
Explicit barycentric weights for polynomial interpolation in the roots or extrema of classical orthogonal polynomials
journal, March 2014
- Wang, Haiyong; Huybrechs, Daan; Vandewalle, Stefan
- Mathematics of Computation, Vol. 83, Issue 290
Roofline: an insightful visual performance model for multicore architectures
journal, April 2009
- Williams, Samuel; Waterman, Andrew; Patterson, David
- Communications of the ACM, Vol. 52, Issue 4
To CG or to HDG: A Comparative Study in 3D
journal, July 2015
- Yakovlev, Sergey; Moxey, David; Kirby, Robert M.
- Journal of Scientific Computing, Vol. 67, Issue 1
Works referencing / citing this record:
Discontinuous Galerkin approximations in computational mechanics: hybridization, exact geometry and degree adaptivity
journal, August 2019
- Giacomini, Matteo; Sevilla, Ruben
- SN Applied Sciences, Vol. 1, Issue 9
hp-adaptive discontinuous Galerkin solver for elliptic equations in numerical relativity
journal, October 2019
- Vincent, Trevor; Pfeiffer, Harald P.; Fischer, Nils L.
- Physical Review D, Vol. 100, Issue 8
Discontinuous Galerkin approximations in computational mechanics: hybridization, exact geometry and degree adaptivity
text, January 2019
- Giacomini, Matteo; Sevilla, Ruben
- arXiv