6 Implementation of Parallel FE Systems
|
|
- Lily Samantha Cannon
- 5 years ago
- Views:
Transcription
1 6 Implementation of Parallel FE Systems 6.1 Implementation of Domain Decomposition in MSC.NASTRAN V Further Parallel Features of MSC.NASTRAN V Parallel Normal Modes Analysis Parallel Direct Frequency Response Analysis 6.3 Hints for Writing Your Own Parallel FEM Programs Components for FEM Subtasks Components for Parallel Solution of Linear Systems 6.4 Questions for Exams
2 6.1 Implementation of DD in MSC.NASTRAN! parallel linear static analysis in MSC.NASTRAN V70.7 will be based on domain decomposition ( parallel SOL101 )! submitted for example by nastran crank mem=20m dmp=4! parallel SOL 101 consists of the following major steps: all processors read the input file (data deck), for example crank.dat containing the complete FE model then all processors execute the automatic partitioning algorithm to create as many domains as there are processors available from then on the processors continue to work on their local domain only each processor performs local element matrix generation, element matrix assembly and constraint elimination all processors cooperate to compute a globally correct solution (for a direct solution of the linear system they have to cooperate in decomposition and FBS) each processor executes a local data recovery if requested by the submittal keyword mergeresults=yes, data is finally collected onto processor 1 ( master )
3 6.1 Implementation of DD in MSC.NASTRAN! parallel SOL 101 is a SPMD program, during its execution each processor executes the following major steps: IFP* start up SEQP EMG EMA UPARTN DCMP (decomposition of interior dofs, chapter 5) DISDCMP (decomposition of boundary dofs) FBS (forward pass on interior dofs) DISFBS (forward and backward pass on boundary dofs) FBS (backward pass on interior dofs) SDR* DISOFPM/S (collection of results) EXIT
4 6.1 Implementation of DD in MSC.NASTRAN! possible design alternative: for example, the model could be split a priori into domains by the preprocessor, so that each processor would be fed by a local input file containing only the local domain would make IFP parallel SEQP would be obsolete but: number of processors would be fixed, often queuing systems are used in industrial environments where user may want to specify a range of the desired number of processors (example: min. 4, max. 8)! example for results with parallel linear static analysis in MSC.NASTRAN (V70.7 development system) FEM description: GRIDs : 49,932 HEXAs : 39,448 PENTAs : 5,752 DOFs : 148,770 1 load case
5 6.1 Implementation of DD in MSC.NASTRAN! Elapsed times on IBM RS/6000 SP (MSC) with direct solution seconds proc. 2 proc. 4 proc. 8 proc.
6 6.1 Implementation of DD in MSC.NASTRAN! Speedups on 8 processors for selected modules with direct solution startup EMG EMA DCMP FBS (87.8) SDR total
7 6.1 Implementation of DD in MSC.NASTRAN! Maximum local disk space with direct solution 1500 megabytes proc. 2 proc. 4 proc. 8 proc.
8 6.1 Implementation of DD in MSC.NASTRAN! Accumulated maximum local disk space with direct solution 2000 megabytes proc. 2 proc. 4 proc. 8 proc.
9 6.1 Implementation of DD in MSC.NASTRAN! Elapsed times on SUN E6000 with iterative solution seconds proc. 2 proc. 4 proc. 8 proc.
10 6.1 Implementation of DD in MSC.NASTRAN! Optimal memory in SOLVIT for runs with iterative solution on SUN E6000 MB proc. 2 proc. 4 proc. 8 proc.! if there is time: view log files and f04 files for serial/parallel runs
11 6.1 Implementation of DD in MSC.NASTRAN! Parallel linear static analysis useful for large models, for example if local disk is insufficient! Very useful for MSC.CONSTRUCT! Postprocessing of created domains with MSC.PATRAN:
12 6.1 Implementation of DD in MSC.NASTRAN! V70.7 prototype successfully tested on IBM RS/6000 Model 590 workstation cluster with switched fast Ethernet! Dedicated talk about this project at MSC Worldwide Automotive Users Conference in September! results without data recovery (would further increase speed ups): minutes proc. 2 proc. 4 proc. FEM description: GRIDs : 225,885 QUAD4s : 205,848 TRIA3s : 12,955 RODs : 364 BARs : 252 ELAS1s : 516 RBE2s : 588 DOFs : 1,348,348 1 load case
13 6.2 Further Parallel Features of MSC.NASTRAN V70.7! V70.7 will be available in October 1999! Distributed parallel analysis types in MSC.NASTRAN V70.7 parallel linear static analysis parallel normal modes analysis parallel direct frequency response analysis! Based on MPI! Supported platforms (current plan) IBM SUN SGI HP NEC Fujitsu Compaq Intel Windows NT V70.7 supports parallel compute servers (IBM SP, HP V-class, etc.) will also work on selected workstation clusters (IBM, HP, SUN, SGI)
14 Processor 1 Processor Parallel Normal Modes Analysis! normal modes analysis = eigenfrequency analysis (SOL 103 in MSC.NASTRAN)! example: vibration of car bodies due to rotations of engine, bumps on roads, etc.! in linear static analysis: a linear system K u = f has to be solved, for example by a decomposition followed by a FBS! in normal modes analysis: generalized eigenvalue problem has to be solved: K x = λ M x K: stiffness matrix M: mass matrix x: eigenvector! several methods available, most efficient method in FEM today is the Lanczos algorithm and its variants (see literature)! typical task of normal modes analysis: determine all eigenvalues between 0 and 300 Hz! parallelization approach: frequency range is split into segments, each processor computes modes in its segment independently F1 F2
15 6.2.1 Parallel Normal Modes Analysis! frequency bounds can be user given or are computed by MSC.NASTRAN with a heuristic formula! Example: CASA satellite, normal modes analysis between 0 and 150 Hz, 209 modes! automatic frequency distribution on 4 processors: 47 EIGENVALUES FOUND IN DISTRIBUTED SEGMENT # 1 58 EIGENVALUES FOUND IN DISTRIBUTED SEGMENT # 2 49 EIGENVALUES FOUND IN DISTRIBUTED SEGMENT # 3 55 EIGENVALUES FOUND IN DISTRIBUTED SEGMENT # 4! elapsed times on IBM RS/6000 SP (66MHz, POWER2): proc. 2 proc. 4 proc. 8 proc. seconds FEM description: GRIDs : 12,283 QUAD4s : 5,999 TRIA3s : 9,044 ELAS1s : 10,231 BARs : 565 BEAMs : 213 CONM2s : 115 dofs : 65, modes
16 6.2.1 Parallel Normal Modes Analysis! further example: large acoustic analysis of car body With Courtesy of FEM description: (structure) GRIDs : 128,659 QUAD4s : 121,912 TRIA3s : 1,234 BARs : 104 CELAS2s : 1,410 CONM2s : 295 RBE2s : 102 RBE32 : 25 DOFs : 768, eigenmodes FEM description: (air) GRIDS : 7,898 HEXA8s : 3,964 PENTA6s : 1,352 TETRA4s : 5,618 DOFs : 7, eigenmodes 56 GB disk, 1.6 TB I/O transferred VOLVO 850 structure 140 frequency steps air
17 6.2.1 Parallel Normal Modes Analysis! Results with V70.0 (serial) CRAY J90 CRAY T90 HP V2250 / PA8200 / 240 MHz, 16 GB main memory, Ultra SCSI, 16 disks, 4 controllers hours CRAY J90 CRAY T90 V2250! Results with V70.7 (serial and parallel) HP V2250 / PA8200 / 240 MHz, 16 GB main memory, Ultra SCSI, 16 disks, 4 controllers HP V2500 / PA8500 / 440 MHz, 16 GB main memory, fibre channel array (10 disks, 2 controllers) HP N4000 / PA8500 / 360 MHz, 8 GB main memory, fibre channel array (10 disks, 2 controllers) hours serial 4 proc. V2250 V2500 N4000
18 6.2.2 Parallel Direct Frequency Response Analysis! Frequency response analysis computes the responses to oscillatory excitation! example: response of car components to excitations resulting from rotations of engine! in direct frequency response (SOL 108), the equation 2 [ ω M + iωb + K] u( ω) = P( ω) is solved directly for each frequency by DCMP followed by FBS! parallelization is straightforward: frequency range is split among processors, each processor computes responses to its local frequencies without interprocessor communication! SOL 108 also contains parallel data recovery: each processor does local data recovery on its local frequencies, at the end results are collected to master via the new DISOFPM/S modules or left local (dependent on mergeresults=yes/no setting)
19 6.2.2 Parallel Direct Frequency Response Analysis! In a parallel SOL 108 run, each processor performs the following steps: read full input deck build element matrices and assembled matrices for full model eliminate constraints determine local frequency segment compute responses to frequencies in local segment do local data recovery collect results on master or leave them local, in the latter case for example MSC.PATRAN can be used to view all local postprocessing files simultaneously (MSC.PATRAN picks automatically the results from the corresponding local results (=xdb) file)
20 6.2.2 Parallel Direct Frequency Response Analysis! Example: exhaust manifold! Elapsed times on IBM RS/6000 SP (POWER2, 66 MHz), 100 frequency steps minutes proc. 2 proc. 4 proc. 8 proc. FEM description: GRIDs : 10,800 QUADs : 6,305 TRIAs : 337 HEXAs : 1,899 PENTAs : 669 TETRAs : 21 DOFs : 49,309
21 6.2.2 Parallel Direct Frequency Response Analysis! Influence of parallel data recovery on 4 processor speedups (40 frequencies) 4 speedup ser. DR par. DR, 1 xdb par. DR, 4 xdb! XY Plot created from multiple xdb files with MSC.PATRAN
22 6.2.2 Parallel Direct Frequency Response Analysis! Car body, ~240,000 dofs, 100 frequencies! Elapsed times on SGI Origin 2000 (300 MHz R12000 processors, 8 GB memory) Similar picture minutes proc. 2 proc. 4 proc. 8 proc.
23 6.3 Hints for Writing Your Own Parallel FEM Programs Components for FEM Subtasks! major steps: input file reading: can be implemented using a finite state machine model partitioning: public domain software available, for example METIS (www-users.cs.umn.edu/~karypis/metis/metis/metis.shtml) (parallel) element matrix generation: see for example book by Schwarz (FORTRAN-Programme zur Methode der finiten Elemente) (parallel) element matrix assembly: easy to implement (parallel) constraint elimination: easy to implement parallel solution of linear systems: see (parallel) data recovery: easy to implement data collection: also easy to implement! the modular structure of MSC.NASTRAN is an excellent example how to write a (parallel) FEM program
24 6.3.2 Components for Parallel Solution of Linear Systems! PSPASES public domain parallel multifrontal solver requires input matrix and right hand side for example in Rutherford- Boeing format (see later) winter.cs.umn.edu/~mjoshi/pspases! PARASOL project funded by the European Union for developing and evaluating parallel sparse matrix solvers started in 1996, will end in August parallel solvers have been developed which will be put into the public domain PSL_PS by GMD, Germany: parallel iterative solver with multigrid preconditioning PSL_MUMPS by RAL, England, and CERFACS, France: parallel direct solver based on multifrontal method
25 6.3.2 Components for Parallel Solution of Linear Systems PSL_DDM by Parallab, Norway: iterative solver based on domain decomposition PSL_FETI by Onera, France: iterative solver based on domain decomposition! the PARASOL solvers are available as a library, all library routines can be called in own programs using the PARASOL library interface, some routines are: psl_init: initialize PARASOL, e.g. select solver psl_map: compute mapping of data to processors psl_solve: solve the linear system psl_end: end PARASOL! the PARASOL software distribution also offers a PARASOL test driver, which reads data in PARASOL file format and outputs the solution, sample call on IBM RS/6000 SP: ptd -um -y -d/tmp/sm/d_bmw3 bmw3_1.mtx bmw3_1.rhs bmw3_1.sln -mi=1,1 -procs 8 -euilib us -labelio yes
26 6.3.2 Components for Parallel Solution of Linear Systems! PARASOL file format is an application of the Rutherford-Boeing file format allows to store all data which a parallel solver might need in a standardized format (even geometry) example: for this project a new MSC.NASTRAN module has been developed, called PARASOL, which outputs the following data for each domain: <testcase>di.mtx: assembled and constrained local stiffness matrix (PSL_MATRIX) <testcase>di.rhs: assembled and constrained local right hand side (PSL_RHSIDE) <testcase>di.vat: variable types (PSL_VARTYPE) <testcase>di.fet: finite element types (PSL_CELLTYPE) <testcase>di.gcl: grid cell list (lists of grids of each element) (PSL_GRIDCELL) <testcase>di.gnd: grid node list (lists variables for each grid) (PSL_GRIDNODE)
27 6.3.2 Components for Parallel Solution of Linear Systems! PARASOL file format (cont d) <testcase>di.geo: coordinates of grids (PSL_NODEPOS) <testcase>di.ivr: list of local interface variables in local numbering (PSL_INTERVAR) <testcase>di.v2g: mapping of all local variables, including interface variables, to global variables (PSL_VAR2GLOB) example: 2x2 cube, split into 2 domains domain 1 domain 2
28 6.3.2 Components for Parallel Solution of Linear Systems grid, element and variable numbers for domain 1 5 (7,8,9) 6 (10,11,12) 17 (43,44,45) 20 (52,53,54) (1,2,3) 14 (34,35,36) 10 (22,23,24) 16 (40,41,42) 18 (46,47,48) 11 (25,26,27) 19 (49,50,51) 21 (55,56,57) 12 (28,29,30) 13 (31,32,33) 15 (37,38,39) y 3 1 z x 7 (13,14,15) 2 (-) 8 (16,17,18) 3 (1,2,3) 9 (19,20,21) 1 (-)
29 6.3.2 Components for Parallel Solution of Linear Systems mtx file for domain 1 stiffness matrix rsa (3I14) (3I14) (1P,3E25.16E3) BC2-_1D E E E E E E E E E E E E E E E
30 6.3.2 Components for Parallel Solution of Linear Systems rhs file for domain 1 right hand side(s) rhsrd r (1P,1E25.16E3) E E E E E E E E E E E E E BC2-_1D1
31 6.3.2 Components for Parallel Solution of Linear Systems vat file for domain 1 variable types avl i 57 1 (1I4) BC2-_1D1
32 6.3.2 Components for Parallel Solution of Linear Systems fet file for domain 1 finite element types avl i 4 1 (1I4) BC2-_1D1
33 6.3.2 Components for Parallel Solution of Linear Systems gcl file for domain 1 finite element list icvs p (6I10) (6I10) BC2-_1D1
34 6.3.2 Components for Parallel Solution of Linear Systems gnd file for domain 1 node list (number of variables per node) ipts p (1I10) (1I10) BC2-_1D1
35 6.3.2 Components for Parallel Solution of Linear Systems geo file for domain 1 node coordinates geo s r 21 3 (1E25.16) E E E E E E E E E E E E E E E E E E E E E E E E E BC2-_1D1
36 6.3.2 Components for Parallel Solution of Linear Systems ivr file for domain 1 interface variables (PSL_INTERVAR) icvs p (1I10) (1I10) BC2-_1D1
37 6.3.2 Components for Parallel Solution of Linear Systems v2g file for domain 1 mapping of local to global indices (PSL_VAR2GLOB) ipts p (1I10) (1I10) BC2-_1D1
38 6.4 Questions for Exams! describe the 7 steps of the (serial) FEM! comparison direct iterative solution of linear systems! classification of parallel computer architectures! describe 4 types of interconnection networks used in commercial parallel computers today! parallel programming: shared memory distributed memory! FEM parallelization approaches! describe the 7+ steps of parallel FEM based on domain decomposition! brief description of multifrontal method and its parallelization
Second Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering
State of the art distributed parallel computational techniques in industrial finite element analysis Second Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering Ajaccio, France
More informationFull Vehicle Dynamic Analysis using Automated Component Modal Synthesis. Peter Schartz, Parallel Project Manager ClusterWorld Conference June 2003
Full Vehicle Dynamic Analysis using Automated Component Modal Synthesis Peter Schartz, Parallel Project Manager Conference Outline Introduction Background Theory Case Studies Full Vehicle Dynamic Analysis
More informationIndustrial finite element analysis: Evolution and current challenges. Keynote presentation at NAFEMS World Congress Crete, Greece June 16-19, 2009
Industrial finite element analysis: Evolution and current challenges Keynote presentation at NAFEMS World Congress Crete, Greece June 16-19, 2009 Dr. Chief Numerical Analyst Office of Architecture and
More information2 Fundamentals of Serial Linear Algebra
. Direct Solution of Linear Systems.. Gaussian Elimination.. LU Decomposition and FBS..3 Cholesky Decomposition..4 Multifrontal Methods. Iterative Solution of Linear Systems.. Jacobi Method Fundamentals
More informationDistributed Memory Parallel MSC.NASTRAN on an IBM Workstation Cluster at Ford Cologne
Authors: Distributed Memory Parallel MSC.NASTRAN on an IBM Workstation Cluster at Ford Cologne Ford Werke Aktiengesellschaft, Spessartstrasse, 50725 Cologne, Germany: Ulrich Viersbach, Tel. +49-221-9034825,
More informationSCALABLE ALGORITHMS for solving large sparse linear systems of equations
SCALABLE ALGORITHMS for solving large sparse linear systems of equations CONTENTS Sparse direct solvers (multifrontal) Substructuring methods (hybrid solvers) Jacko Koster, Bergen Center for Computational
More informationNormal Modes - Rigid Element Analysis with RBE2 and CONM2
APPENDIX A Normal Modes - Rigid Element Analysis with RBE2 and CONM2 T 1 Z R Y Z X Objectives: Create a geometric representation of a tube. Use the geometry model to define an analysis model comprised
More informationApproaches to Parallel Implementation of the BDDC Method
Approaches to Parallel Implementation of the BDDC Method Jakub Šístek Includes joint work with P. Burda, M. Čertíková, J. Mandel, J. Novotný, B. Sousedík. Institute of Mathematics of the AS CR, Prague
More informationGPU COMPUTING WITH MSC NASTRAN 2013
SESSION TITLE WILL BE COMPLETED BY MSC SOFTWARE GPU COMPUTING WITH MSC NASTRAN 2013 Srinivas Kodiyalam, NVIDIA, Santa Clara, USA THEME Accelerated computing with GPUs SUMMARY Current trends in HPC (High
More informationNormal Modes - Rigid Element Analysis with RBE2 and CONM2
LESSON 16 Normal Modes - Rigid Element Analysis with RBE2 and CONM2 Y Y Z Z X Objectives: Create a geometric representation of a tube. Use the geometry model to define an analysis model comprised of plate
More informationNormal Modes - Rigid Element Analysis with RBE2 and CONM2
APPENDIX A Normal Modes - Rigid Element Analysis with RBE2 and CONM2 T 1 Z R Y Z X Objectives: Create a geometric representation of a tube. Use the geometry model to define an analysis model comprised
More informationModal Analysis of Interpolation Constraint Elements and Concentrated Mass
APPENDIX B Modal Analysis of Interpolation Constraint Elements and Concentrated Mass Y Y Z Z X Objectives: Utilize the analysis model created in a previous exercise. Run an MSC.Nastran modal analysis with
More informationStatic and Normal Mode Analysis of a Space Satellite
LESSON 6 Static and Normal Mode of a Space Satellite Z Y X Objectives: Setup and analyze the satellite model for a normal modes and static analysis.. Learn to modify the default subcase parameters, solution
More informationUsing MSC.Nastran for Explicit FEM Simulations
3. LS-DYNA Anwenderforum, Bamberg 2004 CAE / IT III Using MSC.Nastran for Explicit FEM Simulations Patrick Doelfs, Dr. Ingo Neubauer MSC.Software GmbH, D-81829 München, Patrick.Doelfs@mscsoftware.com Abstract:
More informationLS-DYNA s Linear Solver Development Phase 2: Linear Solution Sequence
LS-DYNA s Linear Solver Development Phase 2: Linear Solution Sequence Allen T. Li 1, Zhe Cui 2, Yun Huang 2 1 Ford Motor Company 2 Livermore Software Technology Corporation Abstract This paper continues
More informationModal Analysis of a Flat Plate
WORKSHOP 1 Modal Analysis of a Flat Plate Objectives Produce a MSC.Nastran input file. Submit the file for analysis in MSC.Nastran. Find the first five natural frequencies and mode shapes of the flat plate.
More informationElastic Stability of a Plate
WORKSHOP PROBLEM 7 Elastic Stability of a Plate Objectives Produce a Nastran input file. Submit the file for analysis in MSC/NASTRAN. Find the first five natural modes of the plate. MSC/NASTRAN 101 Exercise
More informationSimulation Advances. Antenna Applications
Simulation Advances for RF, Microwave and Antenna Applications Presented by Martin Vogel, PhD Application Engineer 1 Overview Advanced Integrated Solver Technologies Finite Arrays with Domain Decomposition
More informationModal Analysis of a Beam (SI Units)
APPENDIX 1a Modal Analysis of a Beam (SI Units) Objectives Perform normal modes analysis of a cantilever beam. Submit the file for analysis in MSC.Nastran. Find the first three natural frequencies and
More informationStatic and Normal Mode Analysis of a Space Satellite
LESSON 6 Static and Normal Mode of a Space Satellite Z Y X Objectives: Set up and analyze the Satellite model for a Normal modes and Static analysis.. Learn to modify the default subcase parameters, solution
More informationReckoning With The Limits Of FEM Analysis
Special reprint from CAD CAM 9-10/2008 Reckoning With The Limits Of FEM Analysis 27. Jahrgang 11,90 N 9-10 September/Oktober 2008 TRENDS - TECHNOLOGIEN - BEST PRACTICE DIGITALE FABRIK: VIRTUELLE PRODUKTION
More informationNX Nastran 11 PARALLEL PROCESSING GUIDE
NX Nastran 11 PARALLEL PROCESSING GUIDE 1 Proprietary & Restricted Rights Notice 2016 Siemens Product Lifecycle Management Software Inc. All Rights Reserved. This software and related documentation are
More informationAutomated Component Modal Synthesis with Parallel Processing. Abstract
Automated Component Modal Synthesis with Parallel Processing Paresh Murthy, Petra Poschmann, Mike Reymond, Peter Schartz, Charles T. Wilson, MSC Software Corporation, 2975 Redhill Ave, Costa Mesa, CA 92626
More informationSolving Large Complex Problems. Efficient and Smart Solutions for Large Models
Solving Large Complex Problems Efficient and Smart Solutions for Large Models 1 ANSYS Structural Mechanics Solutions offers several techniques 2 Current trends in simulation show an increased need for
More informationTAU mesh deformation. Thomas Gerhold
TAU mesh deformation Thomas Gerhold The parallel mesh deformation of the DLR TAU-Code Introduction Mesh deformation method & Parallelization Results & Applications Conclusion & Outlook Introduction CFD
More informationSizing Optimization for Industrial Applications
11 th World Congress on Structural and Multidisciplinary Optimisation 07 th -12 th, June 2015, Sydney Australia Sizing Optimization for Industrial Applications Miguel A.A.S: Matos 1, Peter M. Clausen 2,
More informationImplementation of an integrated efficient parallel multiblock Flow solver
Implementation of an integrated efficient parallel multiblock Flow solver Thomas Bönisch, Panagiotis Adamidis and Roland Rühle adamidis@hlrs.de Outline Introduction to URANUS Why using Multiblock meshes
More informationApplications of Finite Element Model Updating Using Experimental Modal Data*
Applications of Finite Element Model Updating Using Experimental Modal Data* E. Dascotte, Dynamic Design Solutions, eddy.dascotte@femtools.com A general procedure for Finite Element model updating, using
More informationPartitioning Effects on MPI LS-DYNA Performance
Partitioning Effects on MPI LS-DYNA Performance Jeffrey G. Zais IBM 138 Third Street Hudson, WI 5416-1225 zais@us.ibm.com Abbreviations: MPI message-passing interface RISC - reduced instruction set computing
More informationUnder the Hood of Implicit LS-DYNA
4 th European LS-DYNA Users Conference Implicit / New Developments Under the Hood of Implicit LS-DYNA Cleve Ashcraft Roger Grimes Brad Maker May 2003 1 Implicit in LS-DYNA v. 970 LS-DYNA v. 970 has an
More informationA Parallel Implementation of the BDDC Method for Linear Elasticity
A Parallel Implementation of the BDDC Method for Linear Elasticity Jakub Šístek joint work with P. Burda, M. Čertíková, J. Mandel, J. Novotný, B. Sousedík Institute of Mathematics of the AS CR, Prague
More informationHFSS Hybrid Finite Element and Integral Equation Solver for Large Scale Electromagnetic Design and Simulation
HFSS Hybrid Finite Element and Integral Equation Solver for Large Scale Electromagnetic Design and Simulation Laila Salman, PhD Technical Services Specialist laila.salman@ansys.com 1 Agenda Overview of
More informationPACKAGE SPECIFICATION HSL 2013
MC55 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To write a supplementary file in Rutherford-Boeing format. The HSL routine MC54 can be used to write matrices in Rutherford-Boeing format and the HSL routine
More informationTwo main topics: `A posteriori (error) control of FEM/FV discretizations with adaptive meshing strategies' `(Iterative) Solution strategies for huge s
. Trends in processor technology and their impact on Numerics for PDE's S. Turek Institut fur Angewandte Mathematik, Universitat Heidelberg Im Neuenheimer Feld 294, 69120 Heidelberg, Germany http://gaia.iwr.uni-heidelberg.de/~ture
More informationPACKAGE SPECIFICATION HSL 2013
MC56 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To read a file held in Rutherford-Boeing format. This file can contain either a sparse matrix or supplementary data. The HSL routines MC54 and MC55 can be
More informationAdvanced Numerical Techniques for Cluster Computing
Advanced Numerical Techniques for Cluster Computing Presented by Piotr Luszczek http://icl.cs.utk.edu/iter-ref/ Presentation Outline Motivation hardware Dense matrix calculations Sparse direct solvers
More informationInstitute of Mechatronics and Information Systems
EXERCISE 2 Free vibrations of a beam arget Getting familiar with the fundamental issues of free vibrations analysis of elastic medium, with the use of a finite element computation system ANSYS. Program
More informationLeveraging Integrated Concurrent Engineering for vehicle dynamics simulation. Manuel CHENE MSC.Software France
Leveraging Integrated Concurrent Engineering for vehicle dynamics simulation Manuel CHENE MSC.Software France Agenda Challenge of vehicle dynamic simulation: frequency domain coverage necessity for a multi
More informationEngine Gasket Model Instructions
SOL 600 Engine Gasket Model Instructions Demonstrated:! Set up the Model Database! 3D Model Import from a MSC.Nastran BDF! Creation of Groups from Element Properties! Complete the Material Models! Import
More informationIntroduction to Parallel Programming for Multicore/Manycore Clusters Part II-3: Parallel FVM using MPI
Introduction to Parallel Programming for Multi/Many Clusters Part II-3: Parallel FVM using MPI Kengo Nakajima Information Technology Center The University of Tokyo 2 Overview Introduction Local Data Structure
More informationMIMD Overview. Intel Paragon XP/S Overview. XP/S Usage. XP/S Nodes and Interconnection. ! Distributed-memory MIMD multicomputer
MIMD Overview Intel Paragon XP/S Overview! MIMDs in the 1980s and 1990s! Distributed-memory multicomputers! Intel Paragon XP/S! Thinking Machines CM-5! IBM SP2! Distributed-memory multicomputers with hardware
More informationModal Analysis of A Flat Plate using Static Reduction
WORKSHOP PROBLEM 2 Modal Analysis of A Flat Plate using Static Reduction Objectives Reduce the dynamic math model, created in Workshop 1, to one with fewer degrees of freedom. Produce a MSC/NASTRAN input
More informationIntegrating LSTC and MSC.Software Technology for Explicit Dynamics and Fluid-Structure Interaction
5 th European LS-DYNA Users Conference Code Developments Integrating LSTC and MSC.Software Technology for Explicit Dynamics and Fluid-Structure Interaction Author: Erik Plugge, MSC.Software Benelux B.V.
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 informationLINUX. Benchmark problems have been calculated with dierent cluster con- gurations. The results obtained from these experiments are compared to those
Parallel Computing on PC Clusters - An Alternative to Supercomputers for Industrial Applications Michael Eberl 1, Wolfgang Karl 1, Carsten Trinitis 1 and Andreas Blaszczyk 2 1 Technische Universitat Munchen
More informationInterface with FE programs
Page 1 of 47 Interdisciplinary > RFlex > Flexible body Interface Interface with FE programs RecurDyn/RFlex can import FE model from ANSYS, NX/NASTRAN, MSC/NASTRAN and I-DEAS. Figure 1 RecurDyn/RFlex Interface
More informationSpeedup Altair RADIOSS Solvers Using NVIDIA GPU
Innovation Intelligence Speedup Altair RADIOSS Solvers Using NVIDIA GPU Eric LEQUINIOU, HPC Director Hongwei Zhou, Senior Software Developer May 16, 2012 Innovation Intelligence ALTAIR OVERVIEW Altair
More informationVibration Analysis of Shaft in SolidWorks and ANSYS
Saimaa University of Applied Sciences Faculty of Technology Lappeenranta Bachelor of Engineering Mechanical Engineering and Production Technology Benjamin Grunwald Vibration Analysis of Shaft in SolidWorks
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 informationA Test Suite for High-Performance Parallel Java
page 1 A Test Suite for High-Performance Parallel Java Jochem Häuser, Thorsten Ludewig, Roy D. Williams, Ralf Winkelmann, Torsten Gollnick, Sharon Brunett, Jean Muylaert presented at 5th National Symposium
More informationSAMCEF MECANO FlexDyn: Market analysis
SAMCEF MECANO FlexDyn: Market analysis Sebastien GOHY 1 - ASD Competence Center - 2011 SAMCEF Mecano Flexdyn: Market analysis SAMCEF Mecano: Reminder Mecano Structure Classical NL FEM Cf Abaqus, MSC Marc
More informationCray XE6 Performance Workshop
Cray XE6 Performance Workshop Multicore Programming Overview Shared memory systems Basic Concepts in OpenMP Brief history of OpenMP Compiling and running OpenMP programs 2 1 Shared memory systems OpenMP
More informationLinear Static Analysis of a Spring Element (CELAS)
Linear Static Analysis of a Spring Element (CELAS) Objectives: Modify nodal analysis and nodal definition coordinate systems to reference a local coordinate system. Define bar elements connected with a
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 informationAn advanced RBF Morph application: coupled CFD-CSM Aeroelastic Analysis of a Full Aircraft Model and Comparison to Experimental Data
An advanced RBF Morph application: coupled CFD-CSM Aeroelastic Analysis of a Full Aircraft Model and Comparison to Experimental Data Dr. Marco Evangelos Biancolini Tor Vergata University, Rome, Italy Dr.
More informationDirect Transient Response with Base Excitation
WORKSHOP PROBLEM 7 Direct Transient Response with Base Excitation Z Y X Objectives Create a geometric representation of a flat rectangular plate. Use the geometry model to define an analysis model comprised
More informationOptimization to Reduce Automobile Cabin Noise
EngOpt 2008 - International Conference on Engineering Optimization Rio de Janeiro, Brazil, 01-05 June 2008. Optimization to Reduce Automobile Cabin Noise Harold Thomas, Dilip Mandal, and Narayanan Pagaldipti
More informationOutline. Parallel Algorithms for Linear Algebra. Number of Processors and Problem Size. Speedup and Efficiency
1 2 Parallel Algorithms for Linear Algebra Richard P. Brent Computer Sciences Laboratory Australian National University Outline Basic concepts Parallel architectures Practical design issues Programming
More informationOPTIMIZATION OF THE CODE OF THE NUMERICAL MAGNETOSHEATH-MAGNETOSPHERE MODEL
Journal of Theoretical and Applied Mechanics, Sofia, 2013, vol. 43, No. 2, pp. 77 82 OPTIMIZATION OF THE CODE OF THE NUMERICAL MAGNETOSHEATH-MAGNETOSPHERE MODEL P. Dobreva Institute of Mechanics, Bulgarian
More informationTFLOP Performance for ANSYS Mechanical
TFLOP Performance for ANSYS Mechanical Dr. Herbert Güttler Engineering GmbH Holunderweg 8 89182 Bernstadt www.microconsult-engineering.de Engineering H. Güttler 19.06.2013 Seite 1 May 2009, Ansys12, 512
More informationEarly Evaluation of the Cray X1 at Oak Ridge National Laboratory
Early Evaluation of the Cray X1 at Oak Ridge National Laboratory Patrick H. Worley Thomas H. Dunigan, Jr. Oak Ridge National Laboratory 45th Cray User Group Conference May 13, 2003 Hyatt on Capital Square
More informationCommodity Cluster Computing
Commodity Cluster Computing Ralf Gruber, EPFL-SIC/CAPA/Swiss-Tx, Lausanne http://capawww.epfl.ch Commodity Cluster Computing 1. Introduction 2. Characterisation of nodes, parallel machines,applications
More informationCost-Effective Parallel Computational Electromagnetic Modeling
Cost-Effective Parallel Computational Electromagnetic Modeling, Tom Cwik {Daniel.S.Katz, cwik}@jpl.nasa.gov Beowulf System at PL (Hyglac) l 16 Pentium Pro PCs, each with 2.5 Gbyte disk, 128 Mbyte memory,
More informationModal Analysis Applications
Modal Analysis and Controls Laboratory Mechanical Engineering Department University of Massachusetts Lowell Presentation Topics Structural Dynamic Modeling Tools MACL Research Overview Correlation Applications
More informationLecture 7: Introduction to HFSS-IE
Lecture 7: Introduction to HFSS-IE 2015.0 Release ANSYS HFSS for Antenna Design 1 2015 ANSYS, Inc. HFSS-IE: Integral Equation Solver Introduction HFSS-IE: Technology An Integral Equation solver technology
More informationMSC.Nastran Structural Optimization Applications for Aerospace Structures
MSC.Nastran Structural Optimization Applications for Aerospace Structures Jack Castro Sr. Technical Representative/Boeing Technical manager Jack Castro Sr. Technical Representative/Boeing Technical manager
More informationGEOMETRY-BASED VIRTUAL MODEL VARIANTS FOR SHAPE OPTIMIZATION AND CAD REFEED
GEOMETRY-BASED VIRTUAL MODEL VARIANTS FOR SHAPE OPTIMIZATION AND CAD REFEED *Dr. Werner Pohl, ** Prof. Dr. Klemens Rother *Fast Concept Modelling & Simulation (FCMS) GmbH, Munich, Germany, **University
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 informationSIMPACKS s FEA Interface New Features in 8.5 and Further Development. Stefan Dietz, INTEC GmbH
SIMPACKS s FEA Interface New Features in 8.5 and Further Development Stefan Dietz, INTEC GmbH Contents New Features in 8.5 integration into SIMPACK GUI generalised file interface between FE-codes and SIMPACK
More informationModal and Stress Analysis of X71A Sport Motorcycle Framebody Virtual Testing Model based on Finite Element Analysis
Modal and Stress Analysis of X71A Sport Motorcycle Framebody Virtual Testing Model based on Finite Element Analysis Andi Wibowo 1, Djoko Setyanto 2 Departement of Mechanical Engineering, Atma Jaya Catholic
More informationRelease Note. Version 5.0r323
Release Note Version 5.0r323 March 3, 2016 Before addressing the new version of AMLS, FastFRS and Delivery Data Base (DDB), CDH will make a number of recommendations based on over 45 years of experience
More informationNormal Modes with Differential Stiffness
WORKSHOP PROBLEM 14b Normal Modes with Differential Stiffness Objectives Analyze a stiffened beam for normal modes. Produce an MSC/ NASTRAN input file that represent beam and load. Submit for analysis.
More informationPATC Parallel Workflows, CSC/PDC
PATC Parallel Workflows, CSC/PDC HPC with Elmer Elmer-team Parallel concept of Elmer MESHING GMSH PARTITIONING ASSEMBLY SOLUTION VISUALIZATION Parallel concept of Elmer MPI Trilinos Pardiso Hypre SuperLU
More informationQLogic TrueScale InfiniBand and Teraflop Simulations
WHITE Paper QLogic TrueScale InfiniBand and Teraflop Simulations For ANSYS Mechanical v12 High Performance Interconnect for ANSYS Computer Aided Engineering Solutions Executive Summary Today s challenging
More informationLS-DYNA s Linear Solver Development Phase1: Element Validation Part II
LS-DYNA s Linear Solver Development Phase1: Element Validation Part II Allen T. Li 1, Zhe Cui 2, Yun Huang 2 1 Ford Motor Company 2 Livermore Software Technology Corporation Abstract This paper continues
More informationD, 8400 E 8400 F
PRODUCT DATA BK Connect Structural Dynamics Correlation Analysis Type 8421 Finite Element Interface Types 8400 D, 8400 E and 8400 F BK Connect Correlation Analysis is an easy to use postprocessing application
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 informationThe Finite Element Method
The Finite Element Method A Practical Course G. R. Liu and S. S. Quek Chapter 1: Computational modeling An overview 1 CONTENTS INTRODUCTION PHYSICAL PROBLEMS IN ENGINEERING COMPUTATIONAL MODELLING USING
More informationParallel Data Mining on a Beowulf Cluster
Parallel Data Mining on a Beowulf Cluster Peter Strazdins, Peter Christen, Ole M. Nielsen and Markus Hegland http://cs.anu.edu.au/ Peter.Strazdins (/seminars) Data Mining Group Australian National University,
More informationDURABILITY ADD-ONS FOR ANSA AND µeta
DURABILITY ADD-ONS FOR ANSA AND µeta Dr. Dietmar Fels Ford Werke GmbH / Germany KEYWORDS Durability, Scripting, Pre and Postprocessing ABSTRACT - The functionality of ANSA and µeta has reached an outstanding
More informationSAToolkit for Nastran (SATK)
SAToolkit for Nastran (SATK) by Carl J. Poplawsky presented to Femap Symposium Ann Arbor, MI. date June 4th, 2015 MAYA Company Overview OEM Foundation Siemens PLM Partner 30+ years Software Developer Femap
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 informationAPPLICATION ON AN UPDATED FINITE ELEMENT MODEL OF AN ENGINE IN THE AUTOMOTIVE INDUSTRY
SISOM 2011 and Session of the Commission of Acoustics, Bucharest 25-26 May APPLICATION ON AN UPDATED FINITE ELEMENT MODEL OF AN ENGINE IN THE AUTOMOTIVE INDUSTRY Gabriel-Petru ANTON, Mihai PAVAL, Fabien
More informationDr. Joe Zhang PDC-3: Parallel Platforms
CSC630/CSC730: arallel & Distributed Computing arallel Computing latforms Chapter 2 (2.3) 1 Content Communication models of Logical organization (a programmer s view) Control structure Communication model
More information2 The Elliptic Test Problem
A Comparative Study of the Parallel Performance of the Blocking and Non-Blocking MPI Communication Commands on an Elliptic Test Problem on the Cluster tara Hafez Tari and Matthias K. Gobbert Department
More informationNormal Modes Analysis of a Simply-Supported Stiffened Plate
APPENDIX C Normal Modes Analysis of a Simply-Supported Stiffened Plate Objectives: Manually convert a Linear Static analysis (Sol 101) input file to a Normal Modes analysis (Sol 103) input file. Learn
More informationRigid Element Analysis with RBAR
WORKSHOP 4 Rigid Element Analysis with RBAR Y Objectives: Idealize the tube with QUAD4 elements. Use RBAR elements to model a rigid end. Produce a Nastran input file that represents the cylinder. Submit
More informationParallel FEM Computation and Multilevel Graph Partitioning Xing Cai
Parallel FEM Computation and Multilevel Graph Partitioning Xing Cai Simula Research Laboratory Overview Parallel FEM computation how? Graph partitioning why? The multilevel approach to GP A numerical example
More information10th August Part One: Introduction to Parallel Computing
Part One: Introduction to Parallel Computing 10th August 2007 Part 1 - Contents Reasons for parallel computing Goals and limitations Criteria for High Performance Computing Overview of parallel computer
More informationCMSC 714 Lecture 6 MPI vs. OpenMP and OpenACC. Guest Lecturer: Sukhyun Song (original slides by Alan Sussman)
CMSC 714 Lecture 6 MPI vs. OpenMP and OpenACC Guest Lecturer: Sukhyun Song (original slides by Alan Sussman) Parallel Programming with Message Passing and Directives 2 MPI + OpenMP Some applications can
More informationModel Solution Memory Tuning from a User s Perspective
Model Solution Memory Tuning from a User s Perspective Pete Ogilvie Product Development Support 1 June 1997 Model Solution and TVM When do you need to worry about this? Setting up your workstation Verifying
More informationStructural Mechanics With ANSYS Pierre THIEFFRY Lead Product Manager ANSYS, Inc.
Structural Mechanics With ANSYS Pierre THIEFFRY Lead Product Manager ANSYS, Inc. 1 ANSYS Structural Mechanics provides highend solver solutions within a highly productive user environment to reliably and
More informationParallel Mesh Partitioning in Alya
Available online at www.prace-ri.eu Partnership for Advanced Computing in Europe Parallel Mesh Partitioning in Alya A. Artigues a *** and G. Houzeaux a* a Barcelona Supercomputing Center ***antoni.artigues@bsc.es
More informationKrishnan Suresh Associate Professor Mechanical Engineering
Large Scale FEA on the GPU Krishnan Suresh Associate Professor Mechanical Engineering High-Performance Trick Computations (i.e., 3.4*1.22): essentially free Memory access determines speed of code Pick
More informationDirect Transient Response Analysis
WORKSHOP 3 Direct Transient Response Analysis Objectives Define time-varying excitation. Produce a MSC.Nastran input file from dynamic math model created in Workshop 1. Submit the file for analysis in
More informationNEi FEA. IRONCAD Advanced FEA. IRONCAD Advanced FEA. NEi FEA
2011 Overview has been designed as a universal, adaptive and user-friendly graphical user interface for geometrical modeling, data input and visualization of results for all types of numerical simulation
More informationEfficient Tridiagonal Solvers for ADI methods and Fluid Simulation
Efficient Tridiagonal Solvers for ADI methods and Fluid Simulation Nikolai Sakharnykh - NVIDIA San Jose Convention Center, San Jose, CA September 21, 2010 Introduction Tridiagonal solvers very popular
More informationGPU Acceleration of Matrix Algebra. Dr. Ronald C. Young Multipath Corporation. fmslib.com
GPU Acceleration of Matrix Algebra Dr. Ronald C. Young Multipath Corporation FMS Performance History Machine Year Flops DEC VAX 1978 97,000 FPS 164 1982 11,000,000 FPS 164-MAX 1985 341,000,000 DEC VAX
More informationParallel Industrial Applications on Windows NT Clusters
Parallel Industrial Applications on Windows NT Clusters K. Takeda 1, N.K. Allsopp 2, J.C. Hardwick 3, P.C. Macey 4, D.A.Nicole 5, S.J.Cox 5 and D.J.Lancaster 5 1 Department of Aeronautics and Astronautics,
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 information