The analysis of computing systems, when performed through simulation, implies in the specification of several aspects of the system, such as its hardware and the running applications. Although there are simulators based in graphical interfaces applied to relatively large systems, the same is not true for the simulation of systems with dozens of thousands of processing elements, such as exascale systems. For such systems current solutions use manual codification of the model to be simulated. Even tough people developing exascale systems can write code without much effort, it is also true that the development can be faster if the process of modeling can be done quickly and automatically. Therefore, our goal is to produce a simulation tool capable of efficient simulation of exascale systems and with a graphical interface for modeling. The research questions to be solved with this project involve two issues, which are the development of an efficient model to parallel simulation, probably using an optimistic approach, and the development of a technique that enable building models with dozens of thousands of elements in a more automatic fashion. The framework for this project is the iSPD (iconic Simulator of Parallel and Distributed Systems), a simulator developed with grants from FAPESP (2008/09312-7 and 2012/15127-3).
News published in Agência FAPESP Newsletter about the scholarship: