Content, progress and pedagogy of the module

Learning objectives

Knowledge

• Must have knowledge about the possible sppedup by using parallelization (Amdahls law / Gustafson-Barsis' law) under different conditions
• Must have knowledge about message and data passing in distributed computing
• Must have knowledge about programming techniques, profiling, benchmarking, code optimization etc.
• Must have knowledge about numerical accuracy in scientific computing problems

Skills

• Must be able to translate the covered principles regarding scientific computing and software development to practice in the programming language(s) utilized in the course
• Must be able to implement software programs to solve scientific computational problems using parallel computing.
• Must be able to implement software programs to solve scientific computational problems using distributed computing units or high-performance specialized computing units (such as GPU)
• Must be able to debug, validate, optimize, benchmark and profile developed software modules.
• Must be able to assess the performance of different hardware architectures for scientific computing problems.

Competences

• Must be able to assess and weigh resources spent on software development against total subsequent computing time for concrete scientific computing problems.
• Must be able to reflect on different software development methods and independently select and combine elements thereof for use in concrete scientific computing problems.

Type of instruction

As described in §17

Exam

Exams