Evaluación del potencial de aceleración de un código profesional de simulación de estructuras de solidificación en aleaciones metálicas por medio de paralelización en CPU y/o GPU

The field of simulation of the physical treatment of materials has been very important for the development of technologies and industrial products for many years. However, the large computational effort required often continues to represent a notable limitation when it comes to being able to take advantage of these simulations in a production environment, especially due to the necessary calculation time. Due to this, the parallelization of computing, and more specifically the possibilities offered by GPU computing today, allow us to consider its use in critical areas within existing algorithms as a possible help when optimizing calculation times. To carry out this work, a professional code has been chosen for the simulation of the growth of grain structures during the solidification of metallic alloys. This code is among the most advanced in its sector and is currently being used by various research centers internationally. The developers have allowed access to it in order to investigate possible improvements through selective parallelization of the code, since, as mentioned above, the current bottleneck is at execution time. Therefore, the objective of this work will be to carry out a thorough investigation of the optimization possibilities of said project, carry out the development of these possible improvements and explain and document both the results and the limitations of the development.