Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Parallel Geometry Inc. proprietary information. Not for redistribution.
prompt 2008-2010 Major Projectsprompt 2008-2010 Major Projects
Redefining the frontiers of computer simulationRedefining the frontiers of computer simulation
Parallel Geometry Inc. proprietary information. Not for redistribution.
Parallel Geometry Inc. proprietary information. Not for redistribution.
Problems in today’s simulations
Lack of integrated spatial data processing chain for
interactive or real-time simulations
• Modeling not closely coupled with simulation
Emergence of Big data in simulations
Critical memory & performance issues in:
• modelling complete engineering systems
• simulating large-scale and/or dynamical systems
Geometric bottlenecks and inadequate computational
models (algorithms) seriously hinder the evolution of
computational scientific
Parallel Geometry Inc. proprietary information. Not for redistribution.
Parallel Geometry Inc. proprietary information. Not for redistribution.
Vision & Mission Statement
Breaking down techno barriers to accuratelyBreaking down techno barriers to accurately
model, simulate and/or control real-world systemsmodel, simulate and/or control real-world systems
Leverage parallel architectures to meet the ‘Great Challenges’ of
science and large-scale computations
Provide a unified and integrated data processing chain for
spatial information
Apply new mathematical methods and algorithms to solve
critical memory & performance issues for large-scale simulations
Enable a new generation of dynamic/adaptative applications
Parallel Geometry Inc. proprietary information. Not for redistribution.
HPCS MarketHPCS Market
High-End Computing (HEC) & SimulationHigh-End Computing (HEC) & Simulationsimulation haute performancesimulation haute performance
NumericalNumerical Simulation Simulationsimulation simulation numériquenumérique
HighHigh ProductivityProductivity ComputingComputing (HPC) (HPC)
High-PerformanceHigh-Performance TechnicalTechnical ComputingComputing
High-PerformanceHigh-Performance ComputingComputing & Simulation (HPCS) & Simulation (HPCS)
ModelingModeling & Simulation & Simulationmodélisationmodélisation & simulation & simulation numériquesnumériques
High-EndHigh-End ComputingComputing & & ScientificScientific VisualizationVisualizationcalculscalculs, simulation & , simulation & visualisationvisualisation haute haute
performanceperformance
Simulation-BasedSimulation-Based Engineering Science (SBES) Engineering Science (SBES)Conception, science et Conception, science et ingénierieingénierie
numériquesnumériques
MemoryMemory & & computationallycomputationally intensive simulation intensive simulation
Computer Computer GraphicsGraphicsinformatiqueinformatique graphiquegraphique
calcul scientifiquecalcul scientifique Scientific ComputingScientific Computing
calcul intensifcalcul intensif Numerically-Intensive ComputingNumerically-Intensive Computing
calcul haute performancecalcul haute performance High-Performance Computing (HPC)High-Performance Computing (HPC)
calcul parallèlecalcul parallèle Parallel ComputingParallel Computing
calculcalcul scientifiquescientifique parallèleparallèle SupercomputingSupercomputing
Parallel Geometry Inc. proprietary information. Not for redistribution.
The Pain in the Market
•• ««Over the past few decades, advances in acquisition, modelingOver the past few decades, advances in acquisition, modelingand simulation technologies have resulted in databases of 3Dand simulation technologies have resulted in databases of 3Dmodels that seem to be unlimited in size and complexitymodels that seem to be unlimited in size and complexity.. » »
— David J. Kasik, Technical Fellow of the Boeing Company (Siggraph 2007)
•• ««We are drowning in data.We are drowning in data. » »
— Janette Wing, Assistant Director of the NSF (Business Week, Dec. 2007)
•• « Simulating the world around us is becoming as important as« Simulating the world around us is becoming as important aswet experiments in everything from drug discovery towet experiments in everything from drug discovery toastrophysics, and high-power supercomputers will soon becomeastrophysics, and high-power supercomputers will soon becomea tool as essential as the microscope. »a tool as essential as the microscope. »
— Scientific Computing World: Oct.-Nov. 2007
•• ««Supercomputers require electrical power Supercomputers require electrical power —— a lots of it. That a lots of it. Thatelectrical power must come from somewhere, and it has aelectrical power must come from somewhere, and it has asignificant economic and environment cost.significant economic and environment cost. » »
— The HPC Brick Wall (Supercomputing, May 2007)
•• ««A clear trend in Engineering is towards complete systemA clear trend in Engineering is towards complete systemmodellingmodelling…… The computational power needed to cope with full The computational power needed to cope with fullaircraft simulation and optimization aircraft simulation and optimization — from both an economical from both an economicaland ecological point of view and ecological point of view — is 10 is 10 77-10 -10 88 higher than today higher than today’’sscapability.capability. » »
— International Scientific Panel on Petascale Computing(European HPC Task Force, Jan. 2007)
Parallel Geometry Inc. proprietary information. Not for redistribution.
SCIENTIFIC
SIMULATION
Biology• drug design
• genomics
Environment• global climate
• contaminant
transport
Applied
Physics• radiation transport
• supernovae Engineering• crash testing
• aerodynamics
• robotics
Lasers & Energy• combustion
• thermonuclear fusion
Experiments
controversial
Experiments
dangerous
Experiments prohibited or
impossible
Experiments difficult
to instrument
Experiments
expensive
Simulation is a critical driver of scientific
discovery and industrial innovation
Key Drivers of the HPCS Market
Parallel Geometry Inc. proprietary information. Not for redistribution.
The Solution: New Algorithms &
Hybrid Architectures
•• « « Issues and Challenges in Simulation-basedIssues and Challenges in Simulation-based
Engineering Science (SBES) cannot be defeated simplyEngineering Science (SBES) cannot be defeated simply
by building bigger and faster computers.by building bigger and faster computers. » »— “Breaking through today’s memory and performance limitations” :
Report of the NSF Blue Ribbon Panel on SBES , Feb. 2006
Parallel Geometry Inc. proprietary information. Not for redistribution.
“Greatest Common Denominator” for
Multiphysics, Multiscale Simulation
Geometry Graphics
Physics
Integrated Computing
System
Polynomials
Parallel Geometry Inc. proprietary information. Not for redistribution.
Underground
Maritime
Undersea
Multiscale Modeling & Simulation
Ground
infrastructures
and Terrains
Airspace
Space
CAD / CAM
AEC
CFD
GIS
Robotics
Photo-
grammetry
Data
Acquisition
Life
Sciences
Scientific
Computing
LLG
Unification
LLG
Inter-
operability
Markets
Markets
Parallel Geometry Inc. proprietary information. Not for redistribution.
HPCS Value Chain
Parallel
Geometry
Texas
IBM
Intel
AMD
NVIDIA
IBM
Cray
Bull
HP
NEC
Sun
PUSH
PULL
Unisys
Honeywell
Rockwell
Esterline/
CMC
Electronics
Siemens
General Dynamics
Thales GroupBoeingAirbus
Target Customers
Enterprise
Government
Industry
Research
Parallel Geometry Inc. proprietary information. Not for redistribution.
Parallel Geometry Inc. proprietary
information. Not for redistribution.
Our objectives
•• Enable Real-time interaction with Complex ModelsEnable Real-time interaction with Complex Models
•• Unify Modeling, Simulation & Visualization through a commonUnify Modeling, Simulation & Visualization through a commondescription for Geometry, Graphics and Physicsdescription for Geometry, Graphics and Physics
•• Geometric calculation on any computing deviceGeometric calculation on any computing device
•• Provide the 'missing link' for full interoperability and reuse of modelsProvide the 'missing link' for full interoperability and reuse of modelsand simulationsand simulations
•• High-performance computing (HPC) and simulation "in a High-performance computing (HPC) and simulation "in a boxbox““((‘‘co-simulatorsco-simulators’’))
•• Enable a new generation of dynamic/Enable a new generation of dynamic/adaptativeadaptative applications applications
Couple simulations and measurements in dynamic data-drivenCouple simulations and measurements in dynamic data-drivenapplicationsapplications
Fill the interoperability gaps between simulation systems andFill the interoperability gaps between simulation systems andother critical systems (command & control systems, embeddedother critical systems (command & control systems, embeddedsystemssystems……))
Parallel Geometry Inc. proprietary information. Not for redistribution.
Target Market Segments
• Aerospace
• Defense / Homeland Security
Fullfills defense and security historic needs to solve higher-level problems and optimization
• Advanced Transportation / Robotics
• GeoWeb
• Computational Science & Engineering
Parallel Geometry Inc. proprietary information. Not for redistribution.
Projects of Interest
in Homeland Security
• 3D Facial Recognition in crowded situation
• Robotic intervention in emergency situations
• Protection of critical infrastructures (ports, airports, subwaystations, nuclear plants…)
• 3D City maps, including underground infrastructures… for firstresponders
• Wild Fires : Simulation and remote sensing
• 3D geographic information systems (GIS) for urban infrastructuremanagement, homeland security
• Geometry/topology analyses... in N dimension(line of sight calculations, collision detection, designof homological sensor networks…)
• Representation of 3D solids in exact, non-polygonal geometry…for ballistic tests
Parallel Geometry Inc. proprietary information. Not for redistribution.
Contact information
For more information, please contact
• Denis Akzam, CEO
(514) 678-2189
or
• Nicolae Alecu
VP Aerospace, Defense & Security Programs
(514) 616-5360
Thank you !