Parallel Computing
In scientific computing new approaches are being explored to meet the demand for computing power in the various applications that arise in the investigation of complex problems.
GPU computing is the use of the GPU (graphics processing unit) to perform scientific calculation of general purpose in parallel. GPUs are massively parallel processors specialized in floating point computations, which offer much better performance than the CPU for a particular type of applications: those with data parallelism.
GPUs are not faster processors, but parallel ones, wich can run hundreds of threads concurrently. They also have a much greater bandwidth than the CPU, allowing it to maintain a greater data flow from memory to the computer cores. The challenge is to create programs that take full advantage of these features of the hardware to solve computationally complex problems. This requires adopting a new programming model and a different approach to design algorithms.
High Performance Computing, also called HPC, today offers a great opportunity to make significant progress in the field of simulation of complex phenomena and problems, through the use of GPU Computing. The HPC has received a boost with the advent of novel architectures of processors, memory and GPU, which deliver significant improvements in performance, lower power consumption and much more affordable prices.
The CTI-HPC researchs how to exploit the capabilities of the GPU to accelerate fluid flow simulations based on particle methods, which discretize the fluid without using fixed grids. In addition, it is researching the use of the GPU in a transparent manner in high level programming languages.
The center, always with the aim to disseminate and train more students in the use and development of new technologies, being organized in conjunction with the University of Boston, a meeting called PASI, related to GPU computing. For more information on this event visit the following Link.
CTI-HPC invites you to participate in the course on GPU Computing to be held from August 8th to September 1st.
The course will provide a thorough introduction to this topic, covering the implementation of algorithms for general and scientific purposes, as well as discussing the advantages of using GPUs for high-performance computing in the future. After completing the course, it will be possible for attendants who are interested on the subject to take part on related projects on CTI-HPC, and also to apply for academic exchange opportunities. 
