This part of the “Performance and Micro Architecture” (PMA) course gives an overview of two active research topics: programming heterogeneous systems (e.g. GPUs, Intel Xeon Phi) and auto-tuning techniques. An overview of programming techniques are exposed during the first time slot. The second time slot is dedicated to auto-tuning techniques. These methods aim at helping code efficiency improvement by automating the scan of in optimization search space.
This course is part of the research master’s degree in computer science http://master.irisa.fr/index.php/en/.
- Programming heterogeneous systems
- Auto-tuning techniques
Related Books and Links
- Multicore Programming Practice Guide (http://www.multicore-association.org/workgroup/mpp.php).
- Romain Dolbeau, Guillaume Colin de Verdière, François Bodin, ”One OpenCL to Rule Them All?” MuCoCoS-2013.
- OpenACC http://www.openacc-standard.org
- ETP4HPC, Strategic Research Agenda, 2013, http://www.etp4hpc.eu/publications/key-documents/
- The use of OpenACC and OpenMP Accelerator Directives with the Cray Compilation Environment, J. Beyer, http://on-demand.gputechconf.com/gtc/2013/presentations/S3084-OpenACC-OpenMP-Directives-CCE.pdf
- Early Experiences With The OpenMP Accelerator Model, Liao et al, http://rosecompiler.org/ROSE_ResearchPapers/Liao-OpenMP-Accelerator-Model-2013.pdf