Parallel and Distributed Programming
Syllabus, Master's level, 1TD070
- Code
- 1TD070
- Education cycle
- Second cycle
- Main field(s) of study and in-depth level
- Computational Science A1F, Computer Science A1F, Technology A1F
- Grading system
- Pass with distinction, Pass with credit, Pass, Fail
- Finalised by
- The Faculty Board of Science and Technology, 10 March 2016
- Responsible department
- Department of Information Technology
Entry requirements
120 credits in science/engineering. High Performance Programming, 10 credits (or Low-level Parallel Programming, 5 credits, and Scientific Computing and Calculus, 10 credits, where the latter may be included in the 120 credits).
Learning outcomes
To pass, the student should be able to
- develop programs with distributed parallelism, parallel debugging included;
- construct parallel algorithms, i.e. identify parallelism in a given algorithm and implement it;
- analyse properties such as efficiency, speedup etc., of parallel algorithms;
- analyse performance of parallel algorithms.
Content
Classification of parallel computers: different kinds of memory organisations, processors, networks and program control flow. Different kinds of parallelism. MPI (Message Passing Interface) programming and data partitioning. Parallelisation of fundamental algorithms in numerical linear algebra and scientific computing: matrix-vector multiplication, matrix-matrix multiplication, FFT (Fast Fourier Transform), N-body simulation, graph algorithms.
Instruction
Lectures, computer labs, assignments and project assignments.
Assessment
Assignments and project presented both as written reports and oral presentations.