On 12 of April 2013 a four-day course «CUDA Architecture & Directives OpenACC» successfully ended. The course was conducted by our resource center RC CC along with the company  NVIDIA and training center  Applied Parallel Computing. 

Modern graphics processors (GPU), including those that underlie graphics cards significantly surpass classic central processors (CPU) in performance. On their base  the hybrid computing systems that have unique performance characteristics are constructed.

For this reason, GPU is widely used for resource-intensive computations and CPU + GPU architecture is the basis of the most powerful supercomputers in the world. For GPU programming  the hardware and software CUDA architecture, OpenACC directives and other tools of high-level language are used.

In this four-day course with practical and theoretical classes, participants studied the CUDA architecture and OpenACC directives. Through this, course participants could learn how to use GPUs for general-purpose computation. Practical classes were carried out on hybrid supercomputer of Resource Center "Computer Center SPbU",  which is equipped with NVIDIA Tesla  GPUs.

In total the course was attended by 28 participants from various organizations such as:

• St. Petersburg State University (Faculties of Physics, Mathematics and Mechanics, Applied Mathematics and Control Processes)
• EMC corporation
• Transas company
• Computer Vision Systems company
• Lyceum "Physical-Technical School"