CUDA Programming on NVIDIA GPUs
Yachay Tech University and Yachay EP promote research and the use of High Performance Computing in the country.
August 8, 2018
Urcuquí, Imbabura.- The School of Mathematical and Computational Sciences of Yachay Tech University, in coordination with Yachay EP, are organizing the “CUDA PROGRAMMING ON NVIDIA GPU’s” workshop. The event will be held from August 13 – 17 in Yachay EP facilities, at Innopolis. The purpose of the workshop is to train attendees on the use of the tools that High Performance Computing can offer. This will allow to deepen research in their fields of interest and expand it to other fields that are less explored in the country.
The workshop is aimed at students, academics and researchers that wish to learn to use CUDA to develop applications that will be executed in NVIDIA’s Graphics Processing Units. The workshop will last seven hours a day, which will be divided into three hours of theory and four hours of practice. Practice hours will be done with Yachay EP’s Quinde I supercomputer, which has a GPU’s NVIDIA K80.
The workshop will be taught by Mike Giles, guest of Dr. Andreas Griewank, Dean of the School of Mathematical and Computational Sciences of Yachay Tech University and Professor of Scientific Computing at Oxford University, expert in GPU Computing and who has also been working closely with NVIDIA the last 10 years. The content to be taught in “CUDA PROGRAMMING ON NVIDIA GPU’s” is based on the course that students, postgraduate students and researchers are taught every year at Oxford.
CUDA is a parallel computing platform developed by NVIDIA, which has its own programming model and is applicable in supercomputers. This programming model is useful to solve very complex computational problems through the use of GPUs, which have a higher performance than classic computers. A GPU is suitable to address problems that can be expressed as parallel calculations of data with a high arithmetic intensity. That is, it can accelerate the execution of several algorithms, processing them in parallel.
The CUDA programming model in GPUs has utility for image representation and analysis, the general processing of signals or physical simulation and, even, financial computation or computational biology.
NVIDIA computing tools accelerated by GPUs are booming in the most promising discovery areas of supercomputing. The “Quinde I” Supercomputer will be a fundamental strategic tool for the exploration of these areas in the country. At the end of the workshop, the participants will have learned the basics of the programming model and will be able to expand this knowledge through the study of the examples provided by NVIDIA as part of their SDK (Software Development Kit).
Each participant will be given a username and a password for the development of their daily practices in the Supercomputer. The agenda programmed includes guided tours to the services that Yachay EP provides (CDT, FabLab, Quinde I) and a campus tour at Yachay Tech University. This way, both institutions work together to promote training on supercomputing in the country.
NVIDIA is a multinational corporation specialized in the development of graphic processing units and integrated circuit technologies. Their products work in work stations (nodes of a supercomputer), and in personal computers and mobile devices. It is one of the largest suppliers of graphic processing units in the world and it is closely linked to the video game industry.