Grant: Discovery Grants Program - Individual
University of Toronto
$31,000.00
Compiler support for GPU application accelerators
FY 2010-11
Summary
Graphics Processing Units (GPUs) have emerged as a new platform for accelerating compute-intensive applications. GPUs exist in every desktop and laptop computer. Their main purpose is to provide high-speed data processing for graphics-intensive applications, such as computer games. However, in the last two to three years GPU manufacturers have provided software interfaces to allow GPUs to be used for non-graphics applications. This has effectively made it possible to use GPUs as low-cost supercomputers. Thus, it comes as no surprise that GPUs are being used in many fields of science, engineering a medicine to accelerate compute-intensive applications, such as medical image processing, engineering design and financial modeling.
In spite of their success, GPUs remain difficult to program. Users of GPUs, who are often not computer specialists, must spend considerable time (measured in months) to convert their existing programs into ones that can execute on GPUs and for which good acceleration of computations can be achieved. Indeed, this difficulty of programming GPUs is a significant hurdle to their effective use today.
The goal of our research is to address the programmability hurdle of GPUs. We will do so through compiler support that allows users to simply indicate in their programs how they want the program to be executed on a GPU, leaving it to the compiler to generate code that will execute efficiently on the GPU. We developed a framework to ease the porting of sequential code to GPUs. We will leverage this work and extend it in 3 directions: (1) provide compiler support to ease the process of restructuring code to run better on GPUs; (2) provide compiler automation for applying this code restructuring; and (3) facilitate programming of systems with multiple GPUs. Achieving these goals will make this important and emerging platform for high performance computing more successful and wide-spread, enabling innovations in application areas.
Program
Discovery Grants Program - Individual
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
121615-2010
Name nom
Abdelrahman, Tarek
Department département
Electrical and Computer Engineering, Edward S. Rogers Sr. Department of
Organization id
31
Institution établissement
University of Toronto
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2010
Award amount
31000
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
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2704
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Distributed and parallel processing
Research subject fr
Traitement réparti et simultané
Partie
1
Nb partie
5
Application title
Compiler support for GPU application accelerators
Keyword
NSERC
Application summary
Graphics Processing Units (GPUs) have emerged as a new platform for accelerating compute-intensive applications. GPUs exist in every desktop and laptop computer. Their main purpose is to provide high-speed data processing for graphics-intensive applications, such as computer games. However, in the last two to three years GPU manufacturers have provided software interfaces to allow GPUs to be used for non-graphics applications. This has effectively made it possible to use GPUs as low-cost supercomputers. Thus, it comes as no surprise that GPUs are being used in many fields of science, engineering a medicine to accelerate compute-intensive applications, such as medical image processing, engineering design and financial modeling.
In spite of their success, GPUs remain difficult to program. Users of GPUs, who are often not computer specialists, must spend considerable time (measured in months) to convert their existing programs into ones that can execute on GPUs and for which good acceleration of computations can be achieved. Indeed, this difficulty of programming GPUs is a significant hurdle to their effective use today.
The goal of our research is to address the programmability hurdle of GPUs. We will do so through compiler support that allows users to simply indicate in their programs how they want the program to be executed on a GPU, leaving it to the compiler to generate code that will execute efficiently on the GPU. We developed a framework to ease the porting of sequential code to GPUs. We will leverage this work and extend it in 3 directions: (1) provide compiler support to ease the process of restructuring code to run better on GPUs; (2) provide compiler automation for applying this code restructuring; and (3) facilitate programming of systems with multiple GPUs. Achieving these goals will make this important and emerging platform for high performance computing more successful and wide-spread, enabling innovations in application areas.