FDD3003 Parallella beräkningar: teori - maskinvara - programvara, med särskilt fokus på flerkärnig programmering 7,5 hp

• Introduction to Parallelism
• Performance Considerations
• Parallel Architectures
  • Shared Memory
  • Scalable Multi-Processors
  • Interconnection networks
• Parallel Programming Methods and Techniques
  • Task and Data parallelism
  • Programming for performance
  • Overview of programming techniques
• Multi-Core programming
  • Properties of Multi-Core architectures
  • Pthreads
  • OpenMP
• Future Directions
  • GPU, Accelerators

The overall goal of the course is to give basic knowledge of the theory, hardware, and software approaches to parallel computing. Especially, hardware and software challenges and the interactions between them, as well as exposure to research challenges in this field will be emphasized.

After the course you will be able to:

• understand the properties of different parallel architectures
• reason about the performance of a system
• assess the potential and limitations of parallel processing
• chose between different parallelization techniques
• write parallel programs on multi-core machines.

Programming experience in C or Fortran.

När kurs inte längre ges har student möjlighet att examineras under ytterligare två läsår.

• EXA1 - Examination, 7,5 hp, betygsskala: P, F

Examinator beslutar, baserat på rekommendation från KTH:s handläggare av stöd till studenter med funktionsnedsättning, om eventuell anpassad examination för studenter med dokumenterad, varaktig funktionsnedsättning.

Examinator får medge annan examinationsform vid omexamination av enstaka studenter.

Homework assignments for the first part of the program. Programming assignments and project work for the second half of the program.

Bo Peng

• Vid grupparbete har alla i gruppen ansvar för gruppens arbete.
• Vid examination ska varje student ärligt redovisa hjälp som erhållits och källor som använts.
• Vid muntlig examination ska varje student kunna redogöra för hela uppgiften och hela lösningen.

Target audience are interdisciplinary graduate students and researchers with needs of high performance and parallel computing.