Scalability of Elliptic Solvers in NWP. Weather and Climate- Prediction
|
|
- Robert Walsh
- 5 years ago
- Views:
Transcription
1 Background Scaling results Tensor product geometric multigrid Summary and Outlook 1/21 Scalability of Elliptic Solvers in Numerical Weather and Climate- Prediction Eike Hermann Müller, Robert Scheichl University of Bath AMM03 Met Office Satellite Meeting, Oct 24 th 2012
2 Background Scaling results Tensor product geometric multigrid Summary and Outlook 2/21 NGWCP project Next Generation Weather and Climate Prediction project Selection of numerical algorithms to simulate the atmosphere in weather and climate prediction which take advantage of massively parallel architectures. Develop new dynamical core for the Met Office Unified Model which scales up to cores Substantial increase in global model resolution 25km few km degrees of freedom per atmospheric variable Model runtime 1hour for 5 day forecast Solve elliptic PDE for pressure correction in 1second
3 Background Scaling results Tensor product geometric multigrid Summary and Outlook 3/21 1 Background Elliptic PDE in implicit time stepping Model equation Multigrid solvers 2 Scaling results Massively parallel scaling on Hector 3 Tensor product geometric multigrid Parallel scaling results Weak scaling Strong scaling Implementation in DUNE-Grid 4 Summary and Outlook
4 Background Scaling results Tensor product geometric multigrid Summary and Outlook 4/21 Implicit timestepping Large scale atmospheric flow: Navier Stokes equations Du Dt = 2Ω u 1 p + g + Su ρ Dρ = ρ u,... Dt image source: NASA Implicit time stepping Unconditionally stable Larger integration time step t Solve 3d elliptic PDE for pressure correction π at every time step: (α t) 2 c 2 s (a π ) + bπ = RHS Davies et al. Q J Royal Met Soc, 131 (608): , 2005,... Significant proportion of model runtime
5 Background Scaling results Tensor product geometric multigrid Summary and Outlook 5/21 Does the solver scale and perform? Started by testing the following black box solvers: Distributed and Unified Numerics Environment (DUNE) ISTL Bastian et al. 2008, Blatt and Bastian 2007 & 2008 CG preconditioned with aggregation AMG + ILU0 smoother Hypre Developed at LLNL by U. Maier-Yang, R. Falgout and others CG preconditioned with BoomerAMG Matrix ( + AMG) setup costs? Matrix-free geometric multigrid Hand-written Fortran code based on tensor-product multigrid idea Börm, Hiptmair Numerical Algorithms. 26: DUNE-based code with indirect horizontal-, direct vertical-addressing
6 Background Scaling results Tensor product geometric multigrid Summary and Outlook 6/21 Does the solver scale and perform? Comparison of Multigrid solvers for model equation Weak scaling of total time +AMG setup time all times in seconds # proc # dof AMG (DUNE) BoomerAMG geo MG matrix setup time for AMG solvers
7 Background Scaling results Tensor product geometric multigrid Summary and Outlook 7/21 Model equation Simplified model equation for u π on spherical shell [ ω 2 (2d) + λ 2 1 ( r 2 r 2 )] u + u = RHS r r Dimensional analysis: r [1, 1 + h] with h = H/R earth = 10 2 : ( ) 2 ω 2 cs α t λ 2 R earth (α t) 2 (N 0 ) 2 Acoustic waves: c s 550ms 1 Buoyancy frequency N 0 = 0.018s 1 Off-centering parameter α = 1 2 (fully implicit: α = 1, fully explicit: α = 0)
8 Background Scaling results Tensor product geometric multigrid Summary and Outlook 8/21 Model equation Properties 7 point FV discretisation Vertical grid graded r k = 1 + h(k/n z ) 2 R earth /H 100 λ 2 /h 2 1 Anisotropy ( λ/h x z ) 2 = O( ) Horizontal grid e.g. cubed sphere, icosahedral,... no pole singularity as in lat/lon grid Δx Δz h ω2 / x 2 = const. as t to keep Courant number c s t/ x fixed
9 Background Scaling results Tensor product geometric multigrid Summary and Outlook 9/21 Multigrid solvers Multigrid idea: Eliminate error on all scales Hierachy of grids h, 2h, 4h,... Apply simple smoother on all levels, restrict/prolongate between levels Residual equation on coarser grids A (H) e (H) = r (H) Work on coarse grids is cheap! Algorithmically optimal Cost(MG) = O(n) Robust & parallelisable h A u =b (h) (h) (h) A e =r (2h) (2h) (2h) A e =r (4h) (4h) (4h) A e =r (8h) (8h) (8h)
10 Background Scaling results Tensor product geometric multigrid Summary and Outlook 10/21 Setup Weak scaling 1/6 of cubed sphere grid (have also run on entire sphere) Horizontal partitioning only (atmos. physics) # processors problem size n x 2n x, n y 2n y, n z = 128, p 4p Keep ν = c g t/ x = 8.44 fixed (i.e. t decreases) ω t x, λ 2 = (α t) 2 (N 0 ) 2 All runs carried out on Hector Cray XE6 supercomputer 2816 nodes of 2 AMD Opteron 16-core Interlagos 2.3GHz = 90,122 cores NB explicit scheme requires ν 1
11 Background Scaling results Tensor product geometric multigrid Summary and Outlook 11/21 Weak Scaling Black box AMG solvers: # iterations & time per iteration all times in seconds AMG (DUNE) BoomerAMG # proc # dof # iter t iter eff. # iter t iter eff [00%] [00%] [98%] [100%] [97%] [97%] [94%] [97%] [95%] [97%] [92%] [84%] [92%] [32%] as preconditioner for CG
12 Background Scaling results Tensor product geometric multigrid Summary and Outlook 12/21 Setup costs + Anisotropy AMG has coarse level & matrix setup costs Rotating anisotropy due to vertical grading top z y x bottom coarse #processors fine Grid-aligned anisotropy Operator well-behaved in horizontal direction Tensor-product matrix-free geometric multigrid Börm, Hiptmair Numerical Algorithms. 26:
13 Background Scaling results Tensor product geometric multigrid Summary and Outlook 13/21 Tensor-product multigrid Tensor product operator A = A (r) M (horiz) h Vertical eigenmodes A (r) e (r) = ω j t M (r) e (r) u(r, x) = j + M (r) A (horiz) h [for operator (α ) ] Börm, Hiptmair Numerical Algorithms. 26: n z j=1 Vertical line relaxation (e.g. RB Gauss-Seidel) Semi-coarsening in horizontal direction only 2d multigrid convergence rate ρ (2d) max j Meteorological application on 3d lat-lon grid: { } ρ (horiz) [e (r) ] j u j (x)e (r) (r) j Buckeridge, Cullen, Scheichl and Wlasak Q J Royal Met Soc 137 (657): R P
14 Background Scaling results Tensor product geometric multigrid Summary and Outlook 14/21 Geometric multigrid Implementation 1/6th of cubed sphere grid RB SOR with vertical line relaxation horizontal semi-coarsening Halo exchange after each smoothing step & prolongation N halo = 1 + 2(n presmooth + n postsmooth ) = 5 Overlap communication & calculation collect/distribute coarse grid data when # procs > # columns collect distribute collect distribute collect distribute
15 Background Scaling results Tensor product geometric multigrid Summary and Outlook 15/21 Geometric multigrid Parallel Multigrid: volume/interface ratio decreases on coarser levels Hülsemann et al., Lect. Notes in Comp. Science and Engineering (2005) BUT Problem well conditioned on coarser levels (see talk by John Thuburn yesterday) Horizontal coupling (vertical coupling irrelevant due to exact vertical solve): ω 2 x 2 l = ω2 x l 2 7 2l Reduce number of levels Coarsen to 1 column (MG) Coarsen to 1 column/processor (7 levels, shallow MG) 4 levels (very shallow MG) 1-level method to check robustness
16 Background Scaling results Tensor product geometric multigrid Summary and Outlook 16/21 Weak scaling results Reduced number of multigrid levels all times in seconds standard MG n lev = 7 n lev = 4 # proc # dof # t iter # t iter # t iter [00%] [00%] [00%] [99%] [99%] [99%] [98%] [98%] [99%] [97%] [97%] [98%] [96%] [98%] [97%] [95%] [97%] [97%] [93%] [95%] [97%] as preconditioner for CG
17 Background Scaling results Tensor product geometric multigrid Summary and Outlook 17/21 Strong scaling results Geometric multigrid Problem size: n n 128 parallel efficiency 100% 90% 75% 50% 25% time per iteration [s] # cores efficiency = p 0 T(p 0 ) p T(p) 100% # cores
18 Background Scaling results Tensor product geometric multigrid Summary and Outlook 18/21 Multigrid on arbitrary spherical grids Grid structure Tensor product grid structure 2-sphere } {{ } host grid 1-column } {{ } directly addressed Hide indirect addressing in horizontal direction by work in vertical direction MacDonald et al., Int J of HPC Appl (2011) Naturally maps to DUNE data model: Attach vector of size n z to each cell of the 2d host grid Multigrid hierarchy only on host grid Size of vertical column O(100)
19 Background Scaling results Tensor product geometric multigrid Summary and Outlook 19/21 Comparison to DUNE geometric MG code Time per iteration [Intel(R) Core(TM)2 Duo CPU E GHz] time per iteration [s] ALUGrid SPGrid+GeometryGrid YaspGrid+GeometryGrid Fortran t iter = A(grid) + B n z n z Implemented together with Andreas Dedner (Warwick)
20 Background Scaling results Tensor product geometric multigrid Summary and Outlook 20/21 Spherical grids Parallel convergence history [preliminary] Cubed sphere r / r CubedSphere SOR CubedSphere Jacobi Icosahedral SOR Icosahedral Jacobi cells, 96 cores Icosahedral grid iteration cells, 320 cores
21 Background Scaling results Tensor product geometric multigrid Summary and Outlook 21/21 Summary and outlook Summary Outlook Multigrid solvers for elliptic PDE in NWP implicit time stepping Verified weak & strong scaling to cores (HECToR) Geometric multigrid code avoids AMG- and matrix setup costs Anisotropy: Tensor product multigrid semi-coarsening + vertical line relaxation Problem well-conditioned on coarser grids use small number of multigrid levels Geometric multigrid robust Hybrid MPI+OpenMP parallelisation More realistic problems: non-symmetry, non-smoothness,... GPGPUs memory layout?
22 22/21 Strong scaling Strong scaling AMG (DUNE) Scaled efficiency re. 64 cores (1node = 32 cores on Hector) E(p) = t(64) 64 p t(p) # dof = = # proc data/halo t/iter E(p) data/halo t/iter E / / % 4096/ % / % 1024/ % / % 256/ % / % 64/ %
23 23/21 Anisotropy BoomerAMG scaling, Setup I: on 16 cores ( ) 2 vertical coupling γ horizontal coupling = λ2 z h 2 x all times in seconds # proc γ top γ middle γ bottom time/iteration t setup [00%] 2.6 [00%] [100%] 2.7 [95%] [97%] 2.8 [92%] [97%] 3.2 [81%] [97%] 3.6 [72%] [84%] 5.7 [45%] [32%] 7.1 [37%]
24 24/21 Anisotropy BoomerAMG scaling, Setup II: on 16 cores ( ) 2 vertical coupling γ horizontal coupling = λ2 z h 2 x all times in seconds # proc γ top γ middle γ bottom time/iteration t setup [00%] 0.61 [00%] [97%] 0.66 [93%] [96%] 0.72 [85%] [94%] 0.86 [71%] [93%] 1.2 [52%] [87%] 1.5 [40%] [82%] 3.0 [20%]
25 25/21 Anisotropy BoomerAMG scaling, Setup III: on 16 cores ( ) 2 vertical coupling γ horizontal coupling = λ2 z h 2 x all times in seconds # proc γ top γ middle γ bottom time/iteration t setup [00%] 5.7 [00%] [98%] 6.0 [93%] [96%] 6.2 [92%] [101%] 10.2 [55%] [87%] 11.4 [50%] [41%] 11.7 [48%] [13%] 12.5 [46%]
26 26/21 Implicit timestepping Semi-implicit semi-lagrangian time stepping [F α t G] (n+1) = [F + (1 α) t G] (n) departure point F (n) F (n+1) departure point u (n+1) + α t c p θ (n+1) π (n+1) = R (n) u θ (n+1) = R (n) θ ρ (n+1) + α t ρ (n+1) u (n+1) = R (n) ρ Linearisation π (n+1) = π + π Equation for pressure correction π : (α t) 2 c 2 s (a π ) + bπ = RHS (n)
27 27/21 Robustness Dependence on parameters ω 2 and λ 2 Coefficient of 2nd order term (time step size) ω 2 = ( cs α t R earth ) 2 Use ω 2, 10 ω 2, 100 ω 2 Vertical coupling (depth of atmosphere, buoyancy term) λ 2 /h 2 = ( Rearth ) 2 1 H 1 + (α t) (N 0 ) 2 Use λ 2, 10 2 λ 2, 10 2 λ 2 Cartesian grid in [0, 1] [0, 1] [0, h]
28 28/21 Robustness # of iterations - 1 level method (CG + line smoother) r k / r 0 < 10 3 Iterations (ω 2,λ 2 ) (ω 2,10 2 λ 2 ) (ω 2,10 2 λ 2 ) (10 1 ω 2,λ 2 ) (10 2 ω 2,λ 2 ) [solver does not converge in 200 iterations] (8.4e+06) 64 (3.4e+07) 256 (1.3e+08) 1024 (5.4e+08) 4096 Number of processes (2.1e+09) (8.6e+09) (3.4e+10)
29 29/21 Robustness Number of iterations - Multigrid Iterations (ω 2,λ 2 ) [MG] (ω 2,10 2 λ 2 ) [MG] (ω 2,10 2 λ 2 ) [MG] (10 1 ω 2,λ 2 ) [MG] (10 2 ω 2,λ 2 ) [MG] (ω 2,λ 2 ) [Shallow MG] (ω 2,10 2 λ 2 ) [Shallow MG] (ω 2,10 2 λ 2 ) [Shallow MG] (10 1 ω 2,λ 2 ) [Shallow MG] (10 2 ω 2,λ 2 ) [Shallow MG] (8.4e+06) 64 (3.4e+07) 256 (1.3e+08) 1024 (5.4e+08) 4096 Number of processes (2.1e+09) (8.6e+09) (3.4e+10)
30 30/21 Comparison to DUNE geometric MG code Implemented together with Andreas Dedner (Warwick) Comparison of sequential runtimes ω 2 = , λ 2 /h 2 = dof Time per iteration [Intel(R) Core(TM)2 Duo CPU E GHz] all times in seconds Code Fortran MG DUNE MG rel. difference Cartesian % Spherical % Is n z large enough to hide indirect addressing?
31 31/21 Comparison to DUNE geometric MG code Time per iteration (Cartesian grid) time per iteration [s] ALUGrid SPGrid YaspGrid Fortran t iter = A(grid) + B n z n z
Matrix-free multi-gpu Implementation of Elliptic Solvers for strongly anisotropic PDEs
Iterative Solvers Numerical Results Conclusion and outlook 1/18 Matrix-free multi-gpu Implementation of Elliptic Solvers for strongly anisotropic PDEs Eike Hermann Müller, Robert Scheichl, Eero Vainikko
More informationEfficient multigrid solvers for strongly anisotropic PDEs in atmospheric modelling
Iterative Solvers Numerical Results Conclusion and outlook 1/22 Efficient multigrid solvers for strongly anisotropic PDEs in atmospheric modelling Part II: GPU Implementation and Scaling on Titan Eike
More informationGPU Implementation of Elliptic Solvers in NWP. Numerical Weather- and Climate- Prediction
1/8 GPU Implementation of Elliptic Solvers in Numerical Weather- and Climate- Prediction Eike Hermann Müller, Robert Scheichl Department of Mathematical Sciences EHM, Xu Guo, Sinan Shi and RS: http://arxiv.org/abs/1302.7193
More informationSELECTIVE ALGEBRAIC MULTIGRID IN FOAM-EXTEND
Student Submission for the 5 th OpenFOAM User Conference 2017, Wiesbaden - Germany: SELECTIVE ALGEBRAIC MULTIGRID IN FOAM-EXTEND TESSA UROIĆ Faculty of Mechanical Engineering and Naval Architecture, Ivana
More informationPhD Student. Associate Professor, Co-Director, Center for Computational Earth and Environmental Science. Abdulrahman Manea.
Abdulrahman Manea PhD Student Hamdi Tchelepi Associate Professor, Co-Director, Center for Computational Earth and Environmental Science Energy Resources Engineering Department School of Earth Sciences
More informationHighly Parallel Multigrid Solvers for Multicore and Manycore Processors
Highly Parallel Multigrid Solvers for Multicore and Manycore Processors Oleg Bessonov (B) Institute for Problems in Mechanics of the Russian Academy of Sciences, 101, Vernadsky Avenue, 119526 Moscow, Russia
More informationMultigrid Algorithms for Three-Dimensional RANS Calculations - The SUmb Solver
Multigrid Algorithms for Three-Dimensional RANS Calculations - The SUmb Solver Juan J. Alonso Department of Aeronautics & Astronautics Stanford University CME342 Lecture 14 May 26, 2014 Outline Non-linear
More informationParallel High-Order Geometric Multigrid Methods on Adaptive Meshes for Highly Heterogeneous Nonlinear Stokes Flow Simulations of Earth s Mantle
ICES Student Forum The University of Texas at Austin, USA November 4, 204 Parallel High-Order Geometric Multigrid Methods on Adaptive Meshes for Highly Heterogeneous Nonlinear Stokes Flow Simulations of
More informationETNA Kent State University
Electronic Transactions on Numerical Analysis. Volume, 2, pp. 92. Copyright 2,. ISSN 68-963. ETNA BEHAVIOR OF PLANE RELAXATION METHODS AS MULTIGRID SMOOTHERS IGNACIO M. LLORENTE AND N. DUANE MELSON Abstract.
More informationIntroduction to Multigrid and its Parallelization
Introduction to Multigrid and its Parallelization! Thomas D. Economon Lecture 14a May 28, 2014 Announcements 2 HW 1 & 2 have been returned. Any questions? Final projects are due June 11, 5 pm. If you are
More informationsmooth coefficients H. Köstler, U. Rüde
A robust multigrid solver for the optical flow problem with non- smooth coefficients H. Köstler, U. Rüde Overview Optical Flow Problem Data term and various regularizers A Robust Multigrid Solver Galerkin
More informationACCELERATING CFD AND RESERVOIR SIMULATIONS WITH ALGEBRAIC MULTI GRID Chris Gottbrath, Nov 2016
ACCELERATING CFD AND RESERVOIR SIMULATIONS WITH ALGEBRAIC MULTI GRID Chris Gottbrath, Nov 2016 Challenges What is Algebraic Multi-Grid (AMG)? AGENDA Why use AMG? When to use AMG? NVIDIA AmgX Results 2
More informationREDUCING COMPLEXITY IN PARALLEL ALGEBRAIC MULTIGRID PRECONDITIONERS
SUBMITTED TO SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, SEPTEMBER 2004 REDUCING COMPLEXITY IN PARALLEL ALGEBRAIC MULTIGRID PRECONDITIONERS HANS DE STERCK, ULRIKE MEIER YANG, AND JEFFREY J. HEYS
More informationD036 Accelerating Reservoir Simulation with GPUs
D036 Accelerating Reservoir Simulation with GPUs K.P. Esler* (Stone Ridge Technology), S. Atan (Marathon Oil Corp.), B. Ramirez (Marathon Oil Corp.) & V. Natoli (Stone Ridge Technology) SUMMARY Over the
More informationAutomatic Generation of Algorithms and Data Structures for Geometric Multigrid. Harald Köstler, Sebastian Kuckuk Siam Parallel Processing 02/21/2014
Automatic Generation of Algorithms and Data Structures for Geometric Multigrid Harald Köstler, Sebastian Kuckuk Siam Parallel Processing 02/21/2014 Introduction Multigrid Goal: Solve a partial differential
More informationMultigrid Methods for Markov Chains
Multigrid Methods for Markov Chains Hans De Sterck Department of Applied Mathematics, University of Waterloo collaborators Killian Miller Department of Applied Mathematics, University of Waterloo, Canada
More informationDistributed NVAMG. Design and Implementation of a Scalable Algebraic Multigrid Framework for a Cluster of GPUs
Distributed NVAMG Design and Implementation of a Scalable Algebraic Multigrid Framework for a Cluster of GPUs Istvan Reguly (istvan.reguly at oerc.ox.ac.uk) Oxford e-research Centre NVIDIA Summer Internship
More informationMultigrid Pattern. I. Problem. II. Driving Forces. III. Solution
Multigrid Pattern I. Problem Problem domain is decomposed into a set of geometric grids, where each element participates in a local computation followed by data exchanges with adjacent neighbors. The grids
More informationAmgX 2.0: Scaling toward CORAL Joe Eaton, November 19, 2015
AmgX 2.0: Scaling toward CORAL Joe Eaton, November 19, 2015 Agenda Introduction to AmgX Current Capabilities Scaling V2.0 Roadmap for the future 2 AmgX Fast, scalable linear solvers, emphasis on iterative
More informationParallel Computations
Parallel Computations Timo Heister, Clemson University heister@clemson.edu 2015-08-05 deal.ii workshop 2015 2 Introduction Parallel computations with deal.ii: Introduction Applications Parallel, adaptive,
More informationGlobal Numerical Weather Predictions and the semi-lagrangian semi-implicit dynamical 1 / 27. ECMWF forecast model
Global Numerical Weather Predictions and the semi-lagrangian semi-implicit dynamical core of the ECMWF forecast model Michail Diamantakis ECMWF Symposium on Modelling and Computations in Atmospheric Sciences
More informationRecent developments for the multigrid scheme of the DLR TAU-Code
www.dlr.de Chart 1 > 21st NIA CFD Seminar > Axel Schwöppe Recent development s for the multigrid scheme of the DLR TAU-Code > Apr 11, 2013 Recent developments for the multigrid scheme of the DLR TAU-Code
More informationContents. I The Basic Framework for Stationary Problems 1
page v Preface xiii I The Basic Framework for Stationary Problems 1 1 Some model PDEs 3 1.1 Laplace s equation; elliptic BVPs... 3 1.1.1 Physical experiments modeled by Laplace s equation... 5 1.2 Other
More informationPROGRAMMING OF MULTIGRID METHODS
PROGRAMMING OF MULTIGRID METHODS LONG CHEN In this note, we explain the implementation detail of multigrid methods. We will use the approach by space decomposition and subspace correction method; see Chapter:
More informationMultigrid at Extreme scales: Communication Reducing Data Models and Asynchronous Algorithms. Mark Adams Columbia University
Multigrid at Extreme scales: Communication Reducing Data Models and Asynchronous Algorithms Mark Adams Columbia University Outline Establish a lower bound on solver complexity Apply ideas to Magnetohydrodynamics
More informationGPU Cluster Computing for FEM
GPU Cluster Computing for FEM Dominik Göddeke Sven H.M. Buijssen, Hilmar Wobker and Stefan Turek Angewandte Mathematik und Numerik TU Dortmund, Germany dominik.goeddeke@math.tu-dortmund.de GPU Computing
More informationRadial Basis Function-Generated Finite Differences (RBF-FD): New Opportunities for Applications in Scientific Computing
Radial Basis Function-Generated Finite Differences (RBF-FD): New Opportunities for Applications in Scientific Computing Natasha Flyer National Center for Atmospheric Research Boulder, CO Meshes vs. Mesh-free
More informationHierarchical Hybrid Grids
Hierarchical Hybrid Grids IDK Summer School 2012 Björn Gmeiner, Ulrich Rüde July, 2012 Contents Mantle convection Hierarchical Hybrid Grids Smoothers Geometric approximation Performance modeling 2 Mantle
More informationNumerical Modelling in Fortran: day 6. Paul Tackley, 2017
Numerical Modelling in Fortran: day 6 Paul Tackley, 2017 Today s Goals 1. Learn about pointers, generic procedures and operators 2. Learn about iterative solvers for boundary value problems, including
More informationFOR P3: A monolithic multigrid FEM solver for fluid structure interaction
FOR 493 - P3: A monolithic multigrid FEM solver for fluid structure interaction Stefan Turek 1 Jaroslav Hron 1,2 Hilmar Wobker 1 Mudassar Razzaq 1 1 Institute of Applied Mathematics, TU Dortmund, Germany
More informationEfficient AMG on Hybrid GPU Clusters. ScicomP Jiri Kraus, Malte Förster, Thomas Brandes, Thomas Soddemann. Fraunhofer SCAI
Efficient AMG on Hybrid GPU Clusters ScicomP 2012 Jiri Kraus, Malte Förster, Thomas Brandes, Thomas Soddemann Fraunhofer SCAI Illustration: Darin McInnis Motivation Sparse iterative solvers benefit from
More informationAn Investigation of Directional-Coarsening And Line-Implicit Smoothing Applied to Agglomeration Multigrid
An Investigation of Directional-Coarsening And Line-Implicit Smoothing Applied to Agglomeration Multigrid J. V. Lassaline Ryerson University 35 Victoria St, Toronto, ON, M5B 2K3, Canada D. W. Zingg University
More informationA mass-conservative version of the semi- Lagrangian semi-implicit HIRLAM using Lagrangian vertical coordinates
A mass-conservative version of the semi- Lagrangian semi-implicit HIRLAM using Lagrangian vertical coordinates Peter Hjort Lauritzen Atmospheric Modeling & Predictability Section National Center for Atmospheric
More informationMultigrid Solvers in CFD. David Emerson. Scientific Computing Department STFC Daresbury Laboratory Daresbury, Warrington, WA4 4AD, UK
Multigrid Solvers in CFD David Emerson Scientific Computing Department STFC Daresbury Laboratory Daresbury, Warrington, WA4 4AD, UK david.emerson@stfc.ac.uk 1 Outline Multigrid: general comments Incompressible
More informationc 2006 Society for Industrial and Applied Mathematics
SIAM J. MATRIX ANAL. APPL. Vol. 27, No. 4, pp. 1019 1039 c 2006 Society for Industrial and Applied Mathematics REDUCING COMPLEXITY IN PARALLEL ALGEBRAIC MULTIGRID PRECONDITIONERS HANS DE STERCK, ULRIKE
More informationReducing Communication Costs Associated with Parallel Algebraic Multigrid
Reducing Communication Costs Associated with Parallel Algebraic Multigrid Amanda Bienz, Luke Olson (Advisor) University of Illinois at Urbana-Champaign Urbana, IL 11 I. PROBLEM AND MOTIVATION Algebraic
More informationWorkshop on Efficient Solvers in Biomedical Applications, Graz, July 2-5, 2012
Workshop on Efficient Solvers in Biomedical Applications, Graz, July 2-5, 2012 This work was performed under the auspices of the U.S. Department of Energy by under contract DE-AC52-07NA27344. Lawrence
More informationResilient geometric finite-element multigrid algorithms using minimised checkpointing
Resilient geometric finite-element multigrid algorithms using minimised checkpointing Dominik Göddeke, Mirco Altenbernd, Dirk Ribbrock Institut für Angewandte Mathematik (LS3) Fakultät für Mathematik TU
More informationMultilevel Methods for Forward and Inverse Ice Sheet Modeling
Multilevel Methods for Forward and Inverse Ice Sheet Modeling Tobin Isaac Institute for Computational Engineering & Sciences The University of Texas at Austin SIAM CSE 2015 Salt Lake City, Utah τ 2 T.
More informationA Comparison of Algebraic Multigrid Preconditioners using Graphics Processing Units and Multi-Core Central Processing Units
A Comparison of Algebraic Multigrid Preconditioners using Graphics Processing Units and Multi-Core Central Processing Units Markus Wagner, Karl Rupp,2, Josef Weinbub Institute for Microelectronics, TU
More informationAccelerated ANSYS Fluent: Algebraic Multigrid on a GPU. Robert Strzodka NVAMG Project Lead
Accelerated ANSYS Fluent: Algebraic Multigrid on a GPU Robert Strzodka NVAMG Project Lead A Parallel Success Story in Five Steps 2 Step 1: Understand Application ANSYS Fluent Computational Fluid Dynamics
More informationAn explicit and conservative remapping strategy for semi-lagrangian advection
An explicit and conservative remapping strategy for semi-lagrangian advection Sebastian Reich Universität Potsdam, Potsdam, Germany January 17, 2007 Abstract A conservative semi-lagrangian advection scheme
More informationUnified Model Performance on the NEC SX-6
Unified Model Performance on the NEC SX-6 Paul Selwood Crown copyright 2004 Page 1 Introduction The Met Office National Weather Service Global and Local Area Climate Prediction (Hadley Centre) Operational
More informationAn added mass partitioned algorithm for rigid bodies and incompressible flows
An added mass partitioned algorithm for rigid bodies and incompressible flows Jeff Banks Rensselaer Polytechnic Institute Overset Grid Symposium Mukilteo, WA October 19, 216 Collaborators Bill Henshaw,
More informationA Hybrid Geometric+Algebraic Multigrid Method with Semi-Iterative Smoothers
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS Numer. Linear Algebra Appl. 013; 00:1 18 Published online in Wiley InterScience www.interscience.wiley.com). A Hybrid Geometric+Algebraic Multigrid Method with
More informationFinite Element Multigrid Solvers for PDE Problems on GPUs and GPU Clusters
Finite Element Multigrid Solvers for PDE Problems on GPUs and GPU Clusters Robert Strzodka Integrative Scientific Computing Max Planck Institut Informatik www.mpi-inf.mpg.de/ ~strzodka Dominik Göddeke
More informationScalable, Hybrid-Parallel Multiscale Methods using DUNE
MÜNSTER Scalable Hybrid-Parallel Multiscale Methods using DUNE R. Milk S. Kaulmann M. Ohlberger December 1st 2014 Outline MÜNSTER Scalable Hybrid-Parallel Multiscale Methods using DUNE 2 /28 Abstraction
More informationThe Immersed Interface Method
The Immersed Interface Method Numerical Solutions of PDEs Involving Interfaces and Irregular Domains Zhiiin Li Kazufumi Ito North Carolina State University Raleigh, North Carolina Society for Industrial
More informationThe Icosahedral Nonhydrostatic (ICON) Model
The Icosahedral Nonhydrostatic (ICON) Model Scalability on Massively Parallel Computer Architectures Florian Prill, DWD + the ICON team 15th ECMWF Workshop on HPC in Meteorology October 2, 2012 ICON =
More informationElmer 8/16/2012. Parallel computing concepts. Parallel Computing. Parallel programming models. Parallel computers. Execution model
Parallel computing concepts Elmer Parallel Computing ElmerTeam Parallel computation means executing tasks concurrently A task encapsulates a sequential program and local data, and its interface to its
More informationHigh Performance Computing for PDE Towards Petascale Computing
High Performance Computing for PDE Towards Petascale Computing S. Turek, D. Göddeke with support by: Chr. Becker, S. Buijssen, M. Grajewski, H. Wobker Institut für Angewandte Mathematik, Univ. Dortmund
More informationMixed-Precision GPU-Multigrid Solvers with Strong Smoothers and Applications in CFD and CSM
Mixed-Precision GPU-Multigrid Solvers with Strong Smoothers and Applications in CFD and CSM Dominik Göddeke and Robert Strzodka Institut für Angewandte Mathematik (LS3), TU Dortmund Max Planck Institut
More informationFinite Volume Discretization on Irregular Voronoi Grids
Finite Volume Discretization on Irregular Voronoi Grids C.Huettig 1, W. Moore 1 1 Hampton University / National Institute of Aerospace Folie 1 The earth and its terrestrial neighbors NASA Colin Rose, Dorling
More informationMultigrid solvers M. M. Sussman sussmanm@math.pitt.edu Office Hours: 11:10AM-12:10PM, Thack 622 May 12 June 19, 2014 1 / 43 Multigrid Geometrical multigrid Introduction Details of GMG Summary Algebraic
More informationCUDA. Fluid simulation Lattice Boltzmann Models Cellular Automata
CUDA Fluid simulation Lattice Boltzmann Models Cellular Automata Please excuse my layout of slides for the remaining part of the talk! Fluid Simulation Navier Stokes equations for incompressible fluids
More informationGPU Acceleration of Unmodified CSM and CFD Solvers
GPU Acceleration of Unmodified CSM and CFD Solvers Dominik Göddeke Sven H.M. Buijssen, Hilmar Wobker and Stefan Turek Angewandte Mathematik und Numerik TU Dortmund, Germany dominik.goeddeke@math.tu-dortmund.de
More informationDune: Crete Introduction to the DUNE Software Library. Heraklion, January 16, Andreas Dedner,
Introduction to the DUNE Software Library Andreas Dedner, Heraklion, January 16, 2013 Department of Mathematics University of Warwick www.warwick.ac.uk/go/dune Intoduction to DUNE? Example Problem: Find
More informationHandling Parallelisation in OpenFOAM
Handling Parallelisation in OpenFOAM Hrvoje Jasak hrvoje.jasak@fsb.hr Faculty of Mechanical Engineering and Naval Architecture University of Zagreb, Croatia Handling Parallelisation in OpenFOAM p. 1 Parallelisation
More informationBlock-asynchronous Multigrid Smoothers for GPU-accelerated Systems
Block-asynchronous Multigrid Smoothers for GPU-accelerated Systems Hartwig Anzt 1, Stanimire Tomov 2, Mark Gates 2, Jack Dongarra 2,3,4, and Vincent Heuveline 1 1 Karlsruhe Institute of Technology (KIT),
More informationDendro: Parallel algorithms for multigrid and AMR methods on 2:1 balanced octrees
Dendro: Parallel algorithms for multigrid and AMR methods on 2:1 balanced octrees Rahul S. Sampath, Santi S. Adavani, Hari Sundar, Ilya Lashuk, and George Biros University of Pennsylvania Abstract In this
More informationA massively parallel solver for discrete Poisson-like problems
A massively parallel solver for discrete Poisson-like problems Yvan Notay and Artem Napov Service de Métrologie Nucléaire Université Libre de Bruxelles (C.P. 165/84) 50, Av. F.D. Roosevelt, B-1050 Brussels,
More informationAn Interface-fitted Mesh Generator and Polytopal Element Methods for Elliptic Interface Problems
An Interface-fitted Mesh Generator and Polytopal Element Methods for Elliptic Interface Problems Long Chen University of California, Irvine chenlong@math.uci.edu Joint work with: Huayi Wei (Xiangtan University),
More informationFast Iterative Solvers for Markov Chains, with Application to Google's PageRank. Hans De Sterck
Fast Iterative Solvers for Markov Chains, with Application to Google's PageRank Hans De Sterck Department of Applied Mathematics University of Waterloo, Ontario, Canada joint work with Steve McCormick,
More informationDendro: Parallel algorithms for multigrid and AMR methods on 2:1 balanced octrees
Dendro: Parallel algorithms for multigrid and AMR methods on 2:1 balanced octrees Rahul S. Sampath, Santi S. Adavani, Hari Sundar, Ilya Lashuk, and George Biros Georgia Institute of Technology, Atlanta,
More information3D Helmholtz Krylov Solver Preconditioned by a Shifted Laplace Multigrid Method on Multi-GPUs
3D Helmholtz Krylov Solver Preconditioned by a Shifted Laplace Multigrid Method on Multi-GPUs H. Knibbe, C. W. Oosterlee, C. Vuik Abstract We are focusing on an iterative solver for the three-dimensional
More information14MMFD-34 Parallel Efficiency and Algorithmic Optimality in Reservoir Simulation on GPUs
14MMFD-34 Parallel Efficiency and Algorithmic Optimality in Reservoir Simulation on GPUs K. Esler, D. Dembeck, K. Mukundakrishnan, V. Natoli, J. Shumway and Y. Zhang Stone Ridge Technology, Bel Air, MD
More informationS0432 NEW IDEAS FOR MASSIVELY PARALLEL PRECONDITIONERS
S0432 NEW IDEAS FOR MASSIVELY PARALLEL PRECONDITIONERS John R Appleyard Jeremy D Appleyard Polyhedron Software with acknowledgements to Mark A Wakefield Garf Bowen Schlumberger Outline of Talk Reservoir
More informationIndex. C m (Ω), 141 L 2 (Ω) space, 143 p-th order, 17
Bibliography [1] J. Adams, P. Swarztrauber, and R. Sweet. Fishpack: Efficient Fortran subprograms for the solution of separable elliptic partial differential equations. http://www.netlib.org/fishpack/.
More informationRealistic Animation of Fluids
Realistic Animation of Fluids p. 1/2 Realistic Animation of Fluids Nick Foster and Dimitri Metaxas Realistic Animation of Fluids p. 2/2 Overview Problem Statement Previous Work Navier-Stokes Equations
More informationAn introduction to mesh generation Part IV : elliptic meshing
Elliptic An introduction to mesh generation Part IV : elliptic meshing Department of Civil Engineering, Université catholique de Louvain, Belgium Elliptic Curvilinear Meshes Basic concept A curvilinear
More informationLarge-scale Gas Turbine Simulations on GPU clusters
Large-scale Gas Turbine Simulations on GPU clusters Tobias Brandvik and Graham Pullan Whittle Laboratory University of Cambridge A large-scale simulation Overview PART I: Turbomachinery PART II: Stencil-based
More informationMixed-Precision GPU-Multigrid Solvers with Strong Smoothers and Applications in CFD and CSM
Mixed-Precision GPU-Multigrid Solvers with Strong Smoothers and Applications in CFD and CSM Dominik Göddeke Institut für Angewandte Mathematik (LS3) TU Dortmund dominik.goeddeke@math.tu-dortmund.de SIMTECH
More informationReconstruction of Trees from Laser Scan Data and further Simulation Topics
Reconstruction of Trees from Laser Scan Data and further Simulation Topics Helmholtz-Research Center, Munich Daniel Ritter http://www10.informatik.uni-erlangen.de Overview 1. Introduction of the Chair
More informationAn Introduction to the LFRic Project
An Introduction to the LFRic Project Mike Hobson Acknowledgements: LFRic Project Met Office: Sam Adams, Tommaso Benacchio, Matthew Hambley, Mike Hobson, Chris Maynard, Tom Melvin, Steve Mullerworth, Stephen
More informationA PARALLEL GEOMETRIC MULTIGRID METHOD FOR FINITE ELEMENTS ON OCTREE MESHES
SIAM J. SCI. COMPUT. Vol. 32, No. 3, pp. 1361 1392 c 2010 Society for Industrial and Applied Mathematics A PARALLEL GEOMETRIC MULTIGRID METHOD FOR FINITE ELEMENTS ON OCTREE MESHES RAHUL S. SAMPATH AND
More informationThis is an author-deposited version published in: Eprints ID: 16478
Open Archive TOULOUSE Archive Ouverte (OATAO) OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible. This is an author-deposited
More informationA highly scalable matrix-free multigrid solver for µfe analysis of bone structures based on a pointer-less octree
Talk at SuperCA++, Bansko BG, Apr 23, 2012 1/28 A highly scalable matrix-free multigrid solver for µfe analysis of bone structures based on a pointer-less octree P. Arbenz, C. Flaig ETH Zurich Talk at
More informationComputational Fluid Dynamics and Interactive Visualisation
Computational Fluid Dynamics and Interactive Visualisation Ralf-Peter Mundani 1, Jérôme Frisch 2 1 Computation in Engineering, TUM 2 E3D, RWTH Aachen University Interdisciplinary Cluster Workshop on Visualization
More informationAlgebraic Multigrid (AMG) for Ground Water Flow and Oil Reservoir Simulation
lgebraic Multigrid (MG) for Ground Water Flow and Oil Reservoir Simulation Klaus Stüben, Patrick Delaney 2, Serguei Chmakov 3 Fraunhofer Institute SCI, Klaus.Stueben@scai.fhg.de, St. ugustin, Germany 2
More informationWhy Use the GPU? How to Exploit? New Hardware Features. Sparse Matrix Solvers on the GPU: Conjugate Gradients and Multigrid. Semiconductor trends
Imagine stream processor; Bill Dally, Stanford Connection Machine CM; Thinking Machines Sparse Matrix Solvers on the GPU: Conjugate Gradients and Multigrid Jeffrey Bolz Eitan Grinspun Caltech Ian Farmer
More informationAlgorithms, System and Data Centre Optimisation for Energy Efficient HPC
2015-09-14 Algorithms, System and Data Centre Optimisation for Energy Efficient HPC Vincent Heuveline URZ Computing Centre of Heidelberg University EMCL Engineering Mathematics and Computing Lab 1 Energy
More informationEfficient Finite Element Geometric Multigrid Solvers for Unstructured Grids on GPUs
Efficient Finite Element Geometric Multigrid Solvers for Unstructured Grids on GPUs Markus Geveler, Dirk Ribbrock, Dominik Göddeke, Peter Zajac, Stefan Turek Institut für Angewandte Mathematik TU Dortmund,
More informationAn Efficient, Geometric Multigrid Solver for the Anisotropic Diffusion Equation in Two and Three Dimensions
1 n Efficient, Geometric Multigrid Solver for the nisotropic Diffusion Equation in Two and Three Dimensions Tolga Tasdizen, Ross Whitaker UUSCI-2004-002 Scientific Computing and Imaging Institute University
More informationFOR ALL GRID SIZES. Thor Gjesdal. Christian Michelsen Research A/S. N-5036 Fantoft, Norway SUMMARY
A CELL-CENTERED MULTIGRID ALGORITHM FOR ALL GRID SIZES Thor Gjesdal Christian Michelsen Research A/S N-5036 Fantoft, Norway SUMMARY Multigrid methods are optimal; that is, their rate of convergence is
More informationAn evaluation of the Performance and Scalability of a Yellowstone Test-System in 5 Benchmarks
An evaluation of the Performance and Scalability of a Yellowstone Test-System in 5 Benchmarks WRF Model NASA Parallel Benchmark Intel MPI Bench My own personal benchmark HPC Challenge Benchmark Abstract
More information>> 1 because the resulting discretization matrix is poorly conditioned. Consider the 1D discretization of eqn. 1.1
AN IMPROVED VARIABLE DENSITY PRESSURE PROJECTION SOLVER FOR ADAPTIVE MESHES AUSTEN DUFFY, ALAN KUHNLE, AND MARK SUSSMAN Abstract Tatebe [7] introduced the multigrid preconditioned conjugate gradient method
More informationAn Investigation into Iterative Methods for Solving Elliptic PDE s Andrew M Brown Computer Science/Maths Session (2000/2001)
An Investigation into Iterative Methods for Solving Elliptic PDE s Andrew M Brown Computer Science/Maths Session (000/001) Summary The objectives of this project were as follows: 1) Investigate iterative
More informationAn algebraic multigrid method for isotropic linear elasticity problems on anisotropic meshes
INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN BIOMEDICAL ENGINEERING Int. J. Numer. Meth. Biomed. Engng. 2010; 26:534 553 Published online 10 July 2008 in Wiley InterScience (www.interscience.wiley.com)..1140
More informationEULAG: high-resolution computational model for research of multi-scale geophysical fluid dynamics
Zbigniew P. Piotrowski *,** EULAG: high-resolution computational model for research of multi-scale geophysical fluid dynamics *Geophysical Turbulence Program, National Center for Atmospheric Research,
More informationOpenFOAM: A C++ Library for Complex Physics Simulations
OpenFOAM: A C++ Library for Complex Physics Simulations Hrvoje Jasak Aleksandar Jemcov and Željko Tuković h.jasak@wikki.co.uk Wikki Ltd, United Kingdom FSB, University of Zagreb, Croatia Development Department,
More informationMultigrid Method using OpenMP/MPI Hybrid Parallel Programming Model on Fujitsu FX10
Multigrid Method using OpenMP/MPI Hybrid Parallel Programming Model on Fujitsu FX0 Kengo Nakajima Information Technology enter, The University of Tokyo, Japan November 4 th, 0 Fujitsu Booth S Salt Lake
More informationTowards a complete FEM-based simulation toolkit on GPUs: Geometric Multigrid solvers
Towards a complete FEM-based simulation toolkit on GPUs: Geometric Multigrid solvers Markus Geveler, Dirk Ribbrock, Dominik Göddeke, Peter Zajac, Stefan Turek Institut für Angewandte Mathematik TU Dortmund,
More informationHIGH ACCURACY MULTISCALE MULTIGRID COMPUTATION FOR PARTIAL DIFFERENTIAL EQUATIONS
University of Kentucky UKnowledge University of Kentucky Doctoral Dissertations Graduate School 2010 HIGH ACCURACY MULTISCALE MULTIGRID COMPUTATION FOR PARTIAL DIFFERENTIAL EQUATIONS Yin Wang University
More informationDevelopment and Testing of a Next Generation Spectral Element Model for the US Navy
Development and Testing of a Next Generation Spectral Element Model for the US Navy Alex Reinecke 1, Kevin Viner 1, James Doyle 1, Sasa Gabersek 1, Matus Martini 2, John Mickalakes 3, Dave Ryglicki 4,
More informationThe GPU as a co-processor in FEM-based simulations. Preliminary results. Dipl.-Inform. Dominik Göddeke.
The GPU as a co-processor in FEM-based simulations Preliminary results Dipl.-Inform. Dominik Göddeke dominik.goeddeke@mathematik.uni-dortmund.de Institute of Applied Mathematics University of Dortmund
More informationParallel Adaptive Tsunami Modelling with Triangular Discontinuous Galerkin Schemes
Parallel Adaptive Tsunami Modelling with Triangular Discontinuous Galerkin Schemes Stefan Vater 1 Kaveh Rahnema 2 Jörn Behrens 1 Michael Bader 2 1 Universität Hamburg 2014 PDES Workshop 2 TU München Partial
More informationA 3-D Finite-Volume Nonhydrostatic Icosahedral Model (NIM) Jin Lee
A 3-D Finite-Volume Nonhydrostatic Icosahedral Model (NIM) Jin Lee Earth System Research Laboratory(ESRL) Director Dr. A.E. (Sandy) MacDonald GFDLNSSLARLAOMLGLERLPMEL Aeronomy Lab. Climate Diagnostic center
More informationHigh Performance Computing for PDE Some numerical aspects of Petascale Computing
High Performance Computing for PDE Some numerical aspects of Petascale Computing S. Turek, D. Göddeke with support by: Chr. Becker, S. Buijssen, M. Grajewski, H. Wobker Institut für Angewandte Mathematik,
More informationarxiv: v1 [math.na] 26 Jun 2014
for spectrally accurate wave propagation Vladimir Druskin, Alexander V. Mamonov and Mikhail Zaslavsky, Schlumberger arxiv:406.6923v [math.na] 26 Jun 204 SUMMARY We develop a method for numerical time-domain
More informationA Semi-Lagrangian Discontinuous Galerkin (SLDG) Conservative Transport Scheme on the Cubed-Sphere
A Semi-Lagrangian Discontinuous Galerkin (SLDG) Conservative Transport Scheme on the Cubed-Sphere Ram Nair Computational and Information Systems Laboratory (CISL) National Center for Atmospheric Research
More information