Since the 1960s, the University of Costa Rica (UCR) has ventured into the use of computational tools for solving scientific problems. With the arrival of Matilde on campus, the first electronic computer in Costa Rica, the opportunity arose for new highly specialized careers and professions in computing to be born in the country. Matilde and her heirs were no longer just a scientific tool for the UCR, but they also began to carry out tasks related to the daily operations of the university. The need for knowledge in this area gave rise to the School of Computer Science and Informatics and the Center for Informatics.
Since the late 1990s, if not earlier, the first high-performance computing clusters (HPC) began to be built in the country in an artisanal manner. The idea of an HPC cluster is to connect several high-performance computers to each other via a high-speed network. The way they manage to work together, or in parallel, is by breaking a large problem into several smaller problems, which will be assigned to a single processor, from among the many that may exist in these computers. Thus, the calculation takes much less time than if it were done on a single computer, in addition to allowing for larger and more complex simulations and numerical models. In other situations, it is impossible to use a desktop computer for these monumental tasks. Many of the problems and numerical simulations in science and engineering can benefit from this type of technology.
Multi-institutional endeavor
The creation of an HPC cluster initially came from the Astrophysical Research Laboratory. In the mid-2000s, the Geophysical Research Center CIGEFI did the same, building its own HPC cluster with more powerful features and greater scientific reach. Then the Quantum Chemistry Laboratory of NASA is born, a laboratory created for conducting research through computational chemistry and that houses clusters with CPU and GPU technology. At the same time, the same was happening with the Pattern Recognition and Intelligent Systems Laboratory, PrisLab, and its clusters also in CPU and GPU. Shortly after, in 2016, the Cluster of the Research Center in Science and Engineering of Materials, CICIMA-HPC, began to offer its services. Even so, all these teams have their days numbered, both due to obsolescence and technical failures from age and use.
This is why around 2019 the first steps began to be taken in consolidating HPC resources at the university, with the support of the Computer Center and several faculty members and researchers. The population that needs access to these devices is growing larger every day. More and more research is also starting to use this resource to reach their results. At the same time, new courses are being created that specialize in the use of some HPC scientific tool, both at the undergraduate and graduate levels. We are beginning to observe a scientific and teaching culture that embraces the use of a cluster as a daily tool more naturally. For these reasons, the initiative of the Institutional HPC Cluster was born, to facilitate the generational change of the currently used equipment, while also optimizing resource use and enhancing the university’s position as a regional leader in the HPC field.
A society in constant change
We must not forget that society is in constant change. Now artificial intelligence in all its forms is peeking into our lives. The UCR cannot fall behind, and the HPC Institutional Cluster is another resource that helps us to be at a competitive international level, both for the university and for the country. Tools like this are what make the training of artificial intelligence and machine learning models possible. But it is not enough to just have the tool; it is necessary to train students and teachers in its use, continue supporting research in this direction, and project the impact of this tool on Costa Rican society.
