Pascal Viewer - Graphic Representation of the efficiency Variation of a parallel application analyzed by time markers

Abstract

This work objective is to develop an automated graphing system that receives a formatted JSON output. Through the data collected from the executions of a parallel program with different problem sizes and cores of different processors and calculations realized by the system it is possible to analyze the performance of the program in a faster and clear form. It is intended that users can more quickly analyze the performance of their programs without worrying about manually processing the information. PaScal Viewer, is the beggining of a long trajectory to delivery relevant information to researchers perform as efficient a research as their programs and there is still much to be explored and implemented. This work, guided by Professor Samuel Xavier and co-led by Master Anderson Braulio, has already yielded a publication in The International Conference for High Performance Computing, Networking, Storage, and Analysis, or simply SC, in 2018. And at this conference presentations, tutorials, panels, and discussion forums have included breakthroughs in many areas and inspired new and innovative areas of computing through the years.