Parallel Computing
Location and Time: Mo 9.11.- Mi 11.11. 2009 9-16 350(OMZ) U014, exercises U011The course gives a thorough introduction to parallel programming. Topics covered are
- Overview of the architecture of parallel computers
- Characterisation of parallel algorithms (e.g. Speedup)
- Shared memory programming with OpenMP and Threads
- Distributed memory programming with MPI
- Parallel programming on graphic adapters (CUDA)
Prerequisites: Programming skills in C or C++ (as e.g. acquired in the lecture Informatik I)
Registration: Due to number of computers available the maximal number of participants is 25. Please register by sending an E-Mail to ingrid.hellwig@iwr.uni-heidelberg.de before Wednesday November 4.
Schedule
Time | Monday, Nov. 9 | Tuesday, Nov. 10 | Wednesday, Nov. 11 |
---|---|---|---|
09.15 - 10.45 | Lecture Parallel Hardware | Lecture Basics of Parallel Algorithms and Mesage Passing | Lecture Analysis of Parallel Algorithms |
11.15 - 12.45 | Lecture Shared Memory Concepts and OpenMP | Excercises GNBP - OpenMP | Excercises GNBP - MPI |
14.15 - 15.45 | Excercises Gravitational N-Body Problem (GNBP) | Lecture (R. 532 in INF 368) The Message Passing Interface (MPI) | Lecture Parallelization on Graphics Hardware with CUDA |
16.15 - 17.45 | Excercises GNBP - tiling | Excercises GNBP - MPI | Excercises GNBP - CUDA |
Script
The script is essentially identical to the transparencies, except it is better suited for printing. Printed copies will be provided for the course.Transparencies
TransparenciesExercises
Exercise sheet 1 (pdf) (Monday 14.15 - 17.45 and Tuesday 11.15 - 12.45)- The Gravitational N-Body Problem (Code)
- OpenMP (Code, Solution)
- Tiling (Code, Solution)
- OpenMP+Tiling (Code, Solution)
- Cuda (Code)