Methodologies and tools for performance analysis and optimization

performance models using machine learning (ML) techniques. Characterization study of parallel OpenMP regions through hardware counters. Development of tuning techniques for parallel agent-based modeling (ABM) applications. Optimization of the distribution of agents according to their computational load and communication pattern. Benchmark to evaluate agent-based development platforms.

the software optimization process is often seen as daunting, cumbersome and time-consuming by software developers wishing to fully exploit HPC resources. Therefore, it is necessary to provide new tools and integrate new techniques, such as machine learning, to support these tasks. From the point of view of performance tuning and energy consumption, it continues to be a challenge to define and implement methodologies capable of taking advantage of the potential of the available resources, overcoming the complexity derived from their heterogeneity. Also, we study and develop scalable and efficient methodologies capable of adapting autonomously and intelligently to the complexity of applications and hardware.

Tools and models for the modeling and optimization of the performance of large-scale agent-based applications

Biomedicine, engineering, finance

agent-based applications