Sunday 19th May '13 You are here: Home > People > Jeyan Thiyagalingam
Personal tools

Dr. Jeyan Thiyagalingam

James Martin Fellow

Oxford e-Research Centre
University of Oxford

Tel:  + 44 (0) 1865 610 629
Fax: + 44 (0) 1865 610 612

 

About me


I wanted to be an electronic engineer and did  my  undergraduate degree on Electrical  and Electronic Engineering; although it was interesting, I decided to turn my attention to computing.  I finished my PhD at Imperial College, London specialising on array layouts. After my PhD, I worked in the city before  moving back to academia.  I am also a member of the British Computer Society and Institute of Engineering and Technology (IET and formerly IEE). 


Research Interests

 

My main research interests are  on high performance /parallel computing, parallel algorithms and energy efficient computing.  

  • High Performance Computing
    • Parallel Algorithms
    • Optimisation of large-scale libraries
  • Energy efficient computing
    • Algorithms : The idea is to alter the  algorithms to alter the energy consumption. 
    • Energy Measurement and Profiling Package (EMPPACK)
    • Benchmarks for establishing energy efficiency.
    • Automatic compiler transformations for minimising energy requirements 

I often use the following  applications /problems as my benchmarks:
  • Target tracking and estimation
    • Data association, assignment and auction  problems

    • Image Processing / Segmentation algorithms / automatic image analysis for characterising structural parameters of organs using image analysis

       

      In the past, I have worked on Parallelising compilers  (especially on parallelising array languages to modern architectures such as CUDA) and on the  development of High Level domain specific languages for performance programming.

      Projects

        • Energy Efficient Computing 
        • GPUGen: Automatic Parallelisation of Array Programs to CUDA
        • Parallel MATLAB,  Sample applications developed with the parallel constructs are here

       

       

      Teaching

      • Principles of Parallel Programming -  Parallel MATLAB
      • GPU programming with high-level abstractions
      • Discrete Maths, Object Oriented Programming and Imperative Programming

       

       Recent Publications

       

      • Anne Trefethen, Jeyan Thiyagalingam, Energy-Aware Software: Challenges, Opportunities and Strategies, Journal of Computational Sciences, July 2012 
      • Jeyan Thiyagalingam, Anne Trefethen, Mathematical Libraries and Energy Efficiency, SIAM Conference on Paralle lProcessing for Scientific Computing (SIAM PP12), Savanah, USA, Feb 2012
      • Jeyan Thiyagalingam, Daniel Goodman, Julia Schnabel, Anne Trefethen, Vicente Grau, On the usage of GPU for Efficient Motion Estimation in Medical Image Sequences, Int. Journal of Biomedical Imaging, Accepted, 2011
      • Jeyan Thiyagalingam and Anne Trefethen, Towards Developing Energ Aware Algorithms, MRSC 2011, Bristol, UK, 2011
      • Jing Guo, Wendell Rodrigues,Jeyan Thiyagalingam, Frederic Guyomarch, Sven-Bodo Scholz, Pierre Boulet, Breaking the GPU Programming Barrier with the Auto-Parallelising SAC Compiler, HIPS/IPDPS,  2011
      • Jing Guo, Jeyan Thiyagalingam, Sven-Bodo Scholz, Breaking the GPU Programming Barrier with the Auto-Parallelising SAC Compiler, DAMP 2011
      • Mihai Duta, Jeyan Thiyagalingam, Anne Trefethen, Ayush Goyal, Vicente Grau, Nic Smith, Parallel Simulation for Parameter Estimation of Optical Tissue Properties, EuroPar 2010, European Conference on Parallel Computing, Springer LNCS 6272, 2010
      • Jing Guo, Jeyan Thiyagalingam, Sven-Bodo Scholz, Towards Compiling Sac to CUDA, Trends in Functional Programming, Volume 10, 2009

       

      Previous publications can be found here 

       

      Others

      • Visiting Research Fellow, University of Bristol, Department of Computer Science, Micro-electronics Group.
      • PC Member, SC'12
      • PC Member, ISC'13
      • Technical expert in the European Exascale Software Initiative (EESI2)
      • Member of the iCT/KTN SEWG