In computational materials science and other research domains, there is an increasing demand for computational power, which serves as a strong driving force for high-performance computers. In the early days, higher performance was achieved by designing smaller chips with a higher density of integrated circuits, which followed Moore’s Law, i.e., the performance of chips doubles approximately every two years or 18 months. However, this law failed when we approached the nanometer scale, equivalent to the distance of a few atoms. This scale is already close to the physical limit, so a different solution to this bottleneck has yet to be found. We must change how we increase computational power to meet scientific and routine needs. This is the emergence of supercomputers. Since a single central processing unit (CPU) is insufficient, we can combine multiple CPUs to increase their power. It is not difficult to conceive the idea of building a supercomputer, similar to creating a LEGO toy. Building a supercomputer is a complex process that involves optimizing the topology to connect individual CPUs, cooling systems, and other components.