Parallel Algorithms and Programming (Master 1 MOSIG, 2016-18)

    Presentation

      The use of multiprocessor machines has become mandatory and most applications are now running on machines with parallel processing units, from tablets to supercomputers. These applications concern both public and private scientific research, as well as industrial and commercial enterprises.

      This lecture and lab work presents the parallel computing domain in different aspects: architectural, algorithmic, parallel programming languages, performance measurements, etc. Examples of parallel algorithms in different domains are also studied.

      The course allows you to understand how to design and program a parallel application and run it on a multicore processor or a classical parallel cluster machine. The lab work on machines allows to implement the concepts studied during the lectures, and this with the language extensions ​​OpenMP and MPI.

    Lectures

      Date#Contentslides (pdf 1p/p)slides (pdf 2p/p)
      23/01/181Introduction, seq programs optimisationslides 1pslides 2p
      30/01/182Decompositionslides 1pslides 2p
      Architecture (part. 1)slides 1pslides 2p
      06/02/183Architecture (part. 2)slides 1pslides 2p
      Shared Memory and OpenMP (part . 1)slides 1pslides 2p
      13/02/184Shared Memory and OpenMP (part . 2)slides 1pslides 2p
      Interconnection Networks (part . 1)slides 1pslides 2p
      06/03/185Interconnection Networks (part . 2)slides 1pslides 2p
      Algorithms of collective communicationsslides 1pslides 2p
      13/03/186Message Passing and MPI (T. Ropars)slides 1pslides 2p
      20/03/187Parallel Linear Algebraslides 1pslides 2p
      03/04/188Fault tolerance for Parallel Applications (T. Ropars)slides 1pslides 2p
      10/04/189Algorithms on Rings (contd)slides 1pslides 2p
      Performance evaluationslides 1pslides 2p

      Contacts

      F. DesprezJ.-F. MéhautT. Ropars B. Videau
      Frederic.Desprez (at) inria.frThomas.Ropars (at) imag.fr
      Last update: Mar. 20, 2018