Resch / Roller / Benkert High Performance Computing on Vector Systems 2010

1. Auflage 2010
ISBN: 978-3-642-11851-7
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, 198 Seiten

This book covers the results of the 11th and 12th Tera?op Workshop and continued a series initiated by NEC and the HLRS in 2004. As part of the Tera?op Workbench, it has become a meeting platform for scientists, application developers, international experts and hardware designers to discuss the current state and future directions of supercomputing with the aim of achieving the highest sustained application perf- mance. The Tera?op Workbench Project is a collaboration between the High Perf- mance Computing Center Stuttgart (HLRS) and NEC Deutschland GmbH (NEC HPCE) to support users to achieve their research goals using High Performance Computing. The ?rst stage of the Tera?op Workbench project (2004-2008) c- centrated on user's applications and their optimization for the 72-node NEC SX-8 installation at HLRS. During this stage, numerous individual codes, developed and maintained by researchers or commercial organizations, have been analyzed and - timized. Several of the codes have shown the ability to outreach the TFlop/s thre- old of sustained performance. This created the possibility for new science and a deeper understanding of the underlying physics.

Autoren/Hrsg.

Roller, Sabine

Kobayashi, Hiroaki

Bez, Wolfgang

Galle, Martin

Wang, Xin

Benkert, Katharina

M, Michael

Resch, Michael M.

Sabine,

Inhaltsverzeichnis

1;Preface;4
2;Contents;6
3;Operating System and Software for Large Scale Systems;12
3.1;Light-Weight Kernel with Portals;13
3.1.1;Introduction;13
3.1.2;Kitten;15
3.1.2.1;OFED;16
3.1.2.2;Benchmarks;17
3.1.3;Portals;17
3.1.3.1;Portals: A Brief Introduction;18
3.1.3.2;Optimizing Portals for LWK;19
3.1.3.3;A High-Performance Infiniband NAL;20
3.1.3.4;Benchmarks;21
3.1.4;MPI;22
3.1.4.1;ORTE Job Preparation and Startup;22
3.1.4.2;Job Start on the Light-Weight Kernel;24
3.1.4.3;Architecture of OOB/Portals;25
3.1.5;Conclusion and Future Work;25
3.1.6;References;26
3.2;Towards an Architecture for Management of Very Large Computing Systems;27
3.2.1;Introduction;27
3.2.1.1;Specific Challenges in HPC;28
3.2.2;Challenges;30
3.2.2.1;Jitter;31
3.2.2.2;Scalability;32
3.2.2.3;Data Correlation;33
3.2.2.4;Error Handling;33
3.2.2.5;Scheduler Awareness;34
3.2.2.6;Tool Integration;34
3.2.3;TIMaCS---The Project;35
3.2.3.1;Idea and Objectives;35
3.2.3.2;Issues---Addressed and Not Addressed;36
3.2.3.3;Architecture;37
3.2.4;Administrational Benefits;40
3.2.4.1;Security Constraints;40
3.2.4.2;Mission-Critical Constraints;40
3.2.4.3;Performance Tuning;41
3.2.4.4;Dynamic Changes;41
3.2.4.5;Hardware Management;41
3.2.5;Business-Benefits with Business-Policy Based Management;42
3.2.5.1;Need for Business-Policy Based Job-Scheduling in HPC;42
3.2.5.2;Approach;43
3.2.6;Conclusion and Future Outlook;44
3.2.7;References;44
3.3;Empirical Optimization of Collective Communications with ADCL;46
3.3.1;Introduction and Motivation;46
3.3.2;The Abstract Data and Communication Library (ADCL);47
3.3.3;Semantics of the New ADCL Interfaces;49
3.3.3.1;The Vector-Map Object;49
3.3.3.2;Extension of the ADCL Interfaces;51
3.3.3.3;The New Function Sets;52
3.3.4;Performance Evaluation;53
3.3.4.1;Integration of ADCL;53
3.3.4.2;Setup;55
3.3.4.3;Results;55
3.3.5;Summary and Outlook;57
3.3.6;References;58
4;I/O Strategies;59
4.1;I/O Forwarding on NEC SX-9;60
4.1.1;IOFWD Implementation;60
4.1.1.1;Design of IOFWD;61
4.1.1.2;IOFWD Components;62
4.1.1.3;Implementation Status;63
4.1.1.4;Performance Results;64
4.1.2;IOFWD Usage;64
4.1.2.1;System Overview at HLRS;65
4.1.2.2;Application Workflow Example;66
4.1.2.3;Application Integration, Compilation, Building and Running;67
4.1.2.4;First Real Application Experiences;68
4.1.3;Conclusion;69
4.1.4;References;69
4.2;High-Speed Data Transmission Technology for the NEC SX-9;70
4.2.1;Introduction;70
4.2.2;LSI Technology;72
4.2.2.1;Serial Interface;72
4.2.2.2;Clocks;72
4.2.3;High-Speed Circuit Technology;73
4.2.3.1;Transmission Technology;74
4.2.3.2;Power Noise Countermeasures;76
4.2.4;Summary;78
4.2.5;References;78
5;Grid and Cloud Computing;79
5.1;The Vector Computing Cloud: Toward a Vector Meta-Computing Environment;80
5.1.1;Introduction;81
5.1.2;Basic Concept of the Vector Computing Clouds;82
5.1.3;Prototyping of the Vector Computing Cloud;83
5.1.3.1;Virtualizing Vector Supercomputers: GRID VM for SX;84
5.1.3.2;Job Scheduling on the Vector Computing Cloud;85
5.1.3.3;MPI Environment for Vector Computing Cloud;88
5.1.4;Feasibility Study and Early Performance Evaluations;88
5.1.4.1;Performance Evaluation of the Job Scheduling Mechanism;89
5.1.4.2;System Tests;90
5.1.4.3;Performance of HPL;93
5.1.5;Conclusions;95
5.1.6;References;95
5.2;Full-Scale 3D Vibration Simulator of an Entire Nuclear Power Plant on Simple Orchestration Application Framework;97
5.2.1;Introduction;98
5.2.2;Full-Scale 3D Vibration Simulator for an Entire Nuclear Power Plant;100
5.2.2.1;GDS Application of Full-Scale 3D Vibration Simulator;100
5.2.2.2;Needs of Pipelined Data-Transfer Scenario;101
5.2.3;Development of Simple Orchestration Application Framework;103
5.2.3.1;Functionalities;104
5.2.3.2;Implementation;104
5.2.4;Full-Scale Simulation of High Temperature Test Engineering Reactor;105
5.2.5;Summaries;108
5.2.6;References;109
5.3;Development of Simple Orchestration Application Framework and Its Application to Burning Plasma Simulation;111
5.3.1;Introduction;111
5.3.2;Simple Orchestration Application Framework (SOAF);113
5.3.2.1;Overview of SOAF;113
5.3.2.2;Controller;114
5.3.2.3;Sentinel;115
5.3.2.4;Configuration File;116
5.3.3;Development of Simple Orchestration Application Framework;117
5.3.3.1;Burning Plasma Simulation;118
5.3.4;Experiment;120
5.3.5;Summaries;123
5.3.6;References;124
6;Acoustics and Structural Mechanic;125
6.1;On Sound Generated by a Globally Unstable Round Jet;126
6.1.1;Introduction;126
6.1.2;Numerical Setup;127
6.1.2.1;Round Jet Flow Simulation;127
6.1.2.2;Aeroacoustic Computation;129
6.1.2.3;Parallelization of the Acoustic Solver;131
6.1.3;Results;133
6.1.3.1;Jet Characteristics;133
6.1.3.2;Acoustic Results;135
6.1.4;Conclusion;137
6.1.5;References;137
6.2;Numerical Simulation of Sibilant [s] Using the Real Geometry of a Human Vocal Tract;140
6.2.1;Introduction;140
6.2.1.1;Signal Processing for Consonants;141
6.2.1.2;Sound Induced Flow;141
6.2.1.3;Sibiant [s] in Dental Treatments;141
6.2.1.4;Computational Analyses for Sibilant [s];142
6.2.1.5;Complicated Morphology of Vocal Tracts;142
6.2.1.6;LES and Aeroacoustics;142
6.2.2;Materials and Methods;143
6.2.3;Result;145
6.2.4;Discussion;146
6.2.4.1;Validity of Real Morphological Geometry;147
6.2.4.2;Aeroacoustic Analyses;148
6.2.4.3;Requirements for High Performance Computing;149
6.2.5;Conclusion;150
6.2.6;References;150
6.3;Identification of Anisotropic Elastic Material Properties by Direct Mechanical Simulations: Estimation of Process Chain Resource Requirements;152
6.3.1;Introduction;152
6.3.2;Material and Methods;153
6.3.2.1;Theoretical Background---Standard Mechanics Approach;154
6.3.2.2;Process Chain Description;156
6.3.3;Results;158
6.3.3.1;Single Sub-Domain;158
6.3.3.2;Domain Count;159
6.3.3.3;Accumulation over Process;159
6.3.4;Summary & Conclusions;161
6.3.5;References;162
7;Computational Fluid Dynamics;163
7.1;Downscaling Climate Simulations for Use in Hydrological Modeling of Medium-Sized River Catchments;164
7.1.1;Introduction;164
7.1.2;The CCLM Model;165
7.1.3;Performance on the HLRS Systems;166
7.1.4;Results;167
7.1.5;Discussion;169
7.1.6;References;170
7.2;DNS of Rising Bubbles Using VOF and Balanced Force Surface Tension;172
7.2.1;Introduction;172
7.2.2;Governing Equations;174
7.2.2.1;Continuity and Navier-Stokes Equations;174
7.2.2.2;Interface Tracking by the VOF Method;174
7.2.2.3;Moving Frame of Reference;175
7.2.3;Surface Tension;176
7.2.3.1;The Continuum Surface Force (CSF) Model;176
7.2.3.2;Balanced-Force Algorithm;177
7.2.3.3;Curvature Estimation;178
7.2.4;Numerical Setup;179
7.2.5;Results: Rise Behavior of Bubbles;180
7.2.5.1;Reduction of Spurious Currents;180
7.2.5.2;Terminal Rise Velocity;183
7.2.5.3;Bubble Shape;183
7.2.6;Conclusion;184
7.2.7;References;185
7.3;Large-Eddy Simulation of Double-Row Compound-Angle Film-Cooling: Computational Aspects;186
7.3.1;Introduction;186
7.3.2;Simulation Methods;188
7.3.2.1;Code and Governing Equations;188
7.3.2.2;Integration and Flux Evaluation;189
7.3.2.3;Turbulence Modeling;189
7.3.2.4;Boundary Treatment;190
7.3.2.5;Computational Environment and Measuring Procedure;190
7.3.3;Performance Results and Flow Field Visualization;191
7.3.3.1;Sequential Performance of Individual Code Components;191
7.3.3.2;Overall Parallel Performance;192
7.3.3.3;Flow Field Visualization;195
7.3.4;Conclusions;195
7.3.5;References;196
7.4;Large Eddy Simulation of Wind Turbulence for Appropriate Urban Environment;198