Efficiency Analysis in Code Development for High-Performance Computing Centers

High-Performance Computing (HPC) has become a powerful tool to solve highly complex projects that could not otherwise be addressed. However, developing parallel programs to take advantage of its benefits, making a proper and effective use of the available computing resources, can be a difficult task for beginners or scientists of specific branches of science which have very little to do with computer programming. For this purpose, a transcompiler for automatic parallelization of sequential codes was developed to train beginners in parallel programming and achieve better performances in HPC centers. Current efforts are focused on expanding the transcompiler with an additional module for optimizing sequential and parallel code fragments with the objective of speeding up their execution times. To accomplish this, several techniques for writing efficient code have been evaluated with a view to integrate them into the transcompiler in the future. This manuscript discusses the importance of these techniques to decrease running times of HPC programs. In addition, a set of tests have been developed for analysing and measuring the improvements achieved when applying each of these techniques, all of them with a significant impact on the execution times.