Prof. Alvaro L.G. A. Coutinho, Director, High Performance Computing Center, The Alberto Luis Coimbra for Institute for Graduate Studies and Research in Engineering, The Federal University of Rio de Janeiro, Brazil
Prof. Fernando A. Rochinha, Mechanical Engineering, The Federal University of Rio de Janeiro, Brazil
Prof. Marta L. Q. Mattoso, Computer Science, The Alberto Luis Coimbra for Graduate Studies and Research in Engineering, The Federal University of Rio de Janeiro, Brazil
COPPE – The Alberto Luiz Coimbra Institute for Graduate Studies and Research in Engineering of the Federal University of Rio de Janeiro, Brazil –helped to renew the Brazilian universities; thereby contributing to the country’s overall development. Founded in 1963 by the engineer Alberto Luiz Coimbra, the institution made it possible to create graduate courses in Brazil. Over the course of the last four decades it has become the most important center for engineering research and education in Latin America. COPPE has already awarded more than 12,000 masters and doctoral degrees in its 12 graduate master’s and doctoral courses. In 2013, Nanotechnology Engineering became the 13rd graduate course at COPPE. Presently, the institution has 325 faculty members, 2,800 students and 350 employees. COPPE also has about 124 modern laboratories, which together constitute the country’s largest engineering laboratory complex. COPPE is part of the Center of Technology at the Federal University of Rio de Janeiro.
Based on three distinguishing features – academic excellence; full-time faculty members and students, and commitment to society – COPPE has distinguished itself through its work to increase the understanding and producing highly qualified professionals and innovative teaching methods; thereby serving as a model for other universities and research institutes across the country. The Intel Parallel Computing Center at COPPE, the first in Latin America, will be in its High Performance Computer Center (NACAD). NACAD’s mission is to foster multidisciplinary research on HPC, providing high quality services to COPPE and the Brazilian community. The first project within Intel IPCC’s program is focused on improving seismic imaging for oil and gas.
This is on the forefront of HPC and Parallel Computing, exhausting available resources on large clusters, usually equipped with coprocessors, such as Intel’s Xeon Phi™. Seismic simulations require fine-tuned codes and extensive optimization, given the enormous amount of required processing time and data generated in typical industrial applications.
Recently another complexity layer has been added, the uncertain character of velocity models and parameters of the underlying mathematical models. This has the potential to increase manifold not only the computational effort to generate the uncertain images, but also the data, since several runs will be necessary to sample with a proper method the stochastic space.
Key to the success of this new strategy is highly optimized codes, making use of the utmost techniques in code optimization and tuning, parallel I/O and related enabling technologies on managing the execution of hundreds of jobs on large clusters. Upon successful outcome of this research, we can make a significant impact on the decision-making workflow of Oil and Gas industry.