Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

University of Saskatchewan

$25,000.00

Many-core physically based simulations

FY 2010-11

Summary
Physically-based simulation has made great strides over the last decade, has become an indispensable tool in entertainment applications such as movies and games, and is becoming increasingly important in surgical, emergency, and military training. However, these simulations remain computationally intensive, and if greater realism is to be achieved, larger computational resources need to be leveraged. For the foreseeable future, this means exploiting multi- and many-core (>100 cores) architectures. Existing tools (OpenMP, MPI, CUDA, OpenCL) offer near-term opportunities to incorporate multiple cores into existing codes, but algorithms designed with only a single core in mind will rarely scale up to hundreds of processes. New algorithms must be designed. In physically-based simulation, there has been recent interest in ""reduced-order"" methods, which offer a principled method of replacing high-dimensional, computationally expensive simulations with fast, low-dimensional approximations. Past research has largely explored the methods' ability to trade accuracy for speed by replacing large, sparse matrices, with small, dense matrices. Our recent work has found that these methods can be extended to enable direct control over the key factors necessary for parallelization -- memory bandwidth usage, computational load, and the number of total threads. This suggests that reduced-order methods could be an excellent primitive for many-core physically-based simulations. The proposed research will investigate how this potential can be converted into practical tools, such as libraries that allow game engine designers to create simulations that meet stringent resource budgets, algorithms that automatically detect how to balance simulation computation across an arbitrary number of cores, and procedural methods that allow simulations to be performed at previously impossible resolutions.
Program
Discovery Grants Program - Individual
Location
Saskatchewan, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
386550-2010
Name nom
Kim, Theodore
Department département
Computer Science
Organization id
16
Institution établissement
University of Saskatchewan
Province en
Saskatchewan
Province fr
Saskatchewan
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2010
Award amount
25000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
405
Area of application group en
Environment
Area of application group fr
Environnement
Area of application en
Modelling and mathematical simulation of natural processes
Area of application fr
Modélisation et simulation mathématique des processus naturels
Research subject code
2707
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Graphics
Research subject fr
Infographie
Partie
1
Nb partie
5
Application title
Many-core physically based simulations
Keyword
NSERC
Application summary
Physically-based simulation has made great strides over the last decade, has become an indispensable tool in entertainment applications such as movies and games, and is becoming increasingly important in surgical, emergency, and military training. However, these simulations remain computationally intensive, and if greater realism is to be achieved, larger computational resources need to be leveraged. For the foreseeable future, this means exploiting multi- and many-core (>100 cores) architectures. Existing tools (OpenMP, MPI, CUDA, OpenCL) offer near-term opportunities to incorporate multiple cores into existing codes, but algorithms designed with only a single core in mind will rarely scale up to hundreds of processes. New algorithms must be designed. In physically-based simulation, there has been recent interest in ""reduced-order"" methods, which offer a principled method of replacing high-dimensional, computationally expensive simulations with fast, low-dimensional approximations. Past research has largely explored the methods' ability to trade accuracy for speed by replacing large, sparse matrices, with small, dense matrices. Our recent work has found that these methods can be extended to enable direct control over the key factors necessary for parallelization -- memory bandwidth usage, computational load, and the number of total threads. This suggests that reduced-order methods could be an excellent primitive for many-core physically-based simulations. The proposed research will investigate how this potential can be converted into practical tools, such as libraries that allow game engine designers to create simulations that meet stringent resource budgets, algorithms that automatically detect how to balance simulation computation across an arbitrary number of cores, and procedural methods that allow simulations to be performed at previously impossible resolutions.