Parallel programming technologies on hybrid architectures

Heterogeneous Computations Team HybriLIT (LIT JINR)

Dubna, January 26 - 29, 2016.

The development of effective applications for carrying out computations on heterogeneous computation systems consisting of multi-core CPU, GPU and Intel Xeon Phi coprocessors is discussed.

Lecture and Practical training topics:

  • OpenMP parallel programming technology;
  • MPI parallel programming technology;
  • CUDA parallel programming technology;
  • OpenCL parallel programming technology;
  • Hybrid OpenMP and CUDA parallel programming on multiple GPU computation systems;
  • OpenMP Programming on Intel Xeon Phi coprocessors.
Main steps of parallel applications development with the use of MPI, OpenMP and CUDA, OpenMP+CUDA technologies will be considered on the example of a particular problem solution.

The tutorial is conducted by the heterogeneous computations group of the Laboratory of Information Technologies (LIT JINR). Practical trainings are held on the heterogeneous cluster HybriLIT.
© 2004 Designed by Lyceum of Informational Technologies №1533