“High-performance computing systems and applications” - course 40,000 rubles. from MSU, training 8 weeks. (2 months), Date: November 27, 2023.

Purpose The implementation of the program is basic training for individuals wishing to gain knowledge and skills in the development of high-performance applications for supercomputers.

The program studies the basic concepts and principles of constructing supercomputer systems, the architecture and functionality of supercomputer system software, MPI and OpenMP parallel programming technologies, methods for debugging and tuning the efficiency of parallel programs for supercomputers using MPI message passing technologies and OpenMP.

Particular attention is paid to the study and practical development of the technologies being studied and the methods used building parallel programs using the example of supercomputers installed at Moscow State University: the Lomonosov and Blue systems Gene/P. In particular, the features of using modern compilers for the programming languages ​​C, C++, Fortran installed on these systems are studied, the use parallel mathematical libraries, instrumental systems for debugging and adjusting the efficiency of parallel programs, methods for organizing parallel work with files.

Form of study
Full-time

Scope of the program (academic hours)
Total - 72
Classroom - 60

Admission requirements: persons with higher education, professional activity - scientists in the field of information and communication technologies, mathematical modeling using high-performance computing systems, teachers of higher educational institutions, engineering and technical workers of organizations associated with the use of high-performance computing technology.

Learning outcomes:

Proficient in MPI and OpenMP parallel programming technologies, working with supercomputers and information networks; understanding of the principles of constructing supercomputer architectures and their software, knowledge of basic methods and tools for developing parallel programs for supercomputers.

As a result of mastering the course, the student should:

know:

· fundamentals of methods for developing parallel programs for supercomputers;

· principles of supercomputer architecture design;

· specification of parallel programming technology MPI-1 and MPI-2, composition and features of various open libraries that implement MPI standards;

· OpenMP directives and functions;

· basic concepts and directives of task management systems for supercomputers;

· methods of remote access to supercomputers, ssh and ftp protocols for remote access and copying of information;

· methods for assessing the effectiveness of parallel programs 

be able to:

· effectively apply the studied technologies in the practice of developing parallel programs for the Lomonosov and Blue Gene/P supercomputers.

Own technologies for developing parallel programs for supercomputers.