In the context of current technological advances, the parallel computers are a reality accessible to a large number of users from modest architectures, with several processors on the system, to supercomputers with thousands of processors. The development of efficient parallel algorithms
has become an urgent requirement. There is a strong demand for qualified engineers with specializations related to parallel computing and a solid foundation in electronics, information systems and technology, so as to maintain the pace of development of new hardware and software.
The course syllabus meet these requirements of current developments and evolution of the combined approach of the theoretical and practical solutions to achieve parallel algorithms using the latest technologies and hardware. This provides graduates with appropriate skills and training needs of current scientific qualifications and modern techniques, quality and competitive, allowing them rapid employment after graduation.
Discipline acquaints students with general techniques for developing of the course parallel algorithms (design, performance estimation and analysis of parallel algorithms, parallel algorithms implementing various parallel computing architecture (architectures with shared memory and message passing). It presents basic templates parallelization algorithms (perfect parallelized loops, parallelized loops with dependencies, sequence of loops, multi-level nested loops) and use these templates for designing parallel algorithms fundamentals (vector and matrix operations, reduction, filters applied on images).
Lecturer: Valentin Stoica