The lectures «Programming on multi-processor systems» took place from 2 to 23 April 2013, four two-hour lectures in total. The course was organized by M.I. Polykarpov at ITEP in bld. 180 room 418 (11:00-12:30). The topics of the lectures were aimed to give the listeners basic knowledge in programming on the ITEP FAIR Tier 1 supercomputer (now in construction). The first two lectures were given by the candidate of phys.-math. sciences Ryabkov O.I., the last two-by the candidate of phys.-math. sciences Ulibishev M.V. (MSU, ITEP). The programme of the lectures is down below.
Programme of the lectures «Programming on multi-processor systems»
Ryabkov O.I., «Programming with MPI technology».
April 2, 1st lecture. Basic knowledge on the MPI technology. Comparison with other parallel programming technologies. Simple MPI-program. Launching programs.
April 9, 2nd lecture. Data transfer methods in the MPI technology. Additional MPI abilities. Basic information on the OpenMP technology.
Ulibishev M.V., «Massively parallel calculations on GPUs».
April 16, 3rd lecture. Overview of hardware and software general purpose calculations on GPUs. Basic information on Nvidia CUDA architecture. Comparison of CUDA and OpenCL. Basic information on parallel programming with CUDA C. Examples of simple programs.
April 23, 4th lecture. More complicated programs: usage of constant and shared memory, atomic operations. Optimization and programing on several GPUs.
Photo
