Analysis and optimization of a demographic simulator for parallel environments.

Duration of the project:
January, 2015 –
July, 2015
Collaborators
Project Manager
inLab FIB Team:
Participants
  • Alessandro Pellegrini
Areas of expertise involved in the project
Technology
Analysis and optimization of a demographic simulator for parallel environments.

Description

The objective of this project is to exploit the capabilities offered by the multicore architecture to improve the performance of the socio-demographic dynamics simulator, Yades (Yet Another DEmographic Simulator), created at InLab in collaboration with the Lancaster University and the Barcelona Super Computing Center (BSC).

To achieve this goal, the ROOT-Sim library (ROme OpTimistic Simulator) has been used to take advantage of the benefits offered by a shared memory architecture to optimize the performance of the simulator. This improvement makes it possible to reduce intra-node communications and to provide the simulator with more advanced capabilities to work with a currently highly extended architecture.

To achieve this, a collaboration with the High Performance and Dependable Computing Systems Group (DIAG) of the University of Sapienza was initiated in order to adapt our simulator with its ROOT-Sim library, a general purpose, parallel discrete event simulation (PDES- Parallel Discrete Event Simulation) library. This static library can be used with any model that follows the ANSI-C standard.

This project has not only optimized the performance of the Yades simulator in multicore architecture but has also contributed to the development of some parts of the ROOT-Sim library and its testing with a real application model.

Articles and presentations