Dynamic Memory Allocation Problem
María Soto, André Rossi, Marc Sevaux, Johann Laurent, Narendra Jussien · 2012
This chapter deals with the last version of the dynamic memory allocation problem, called MemExplorer-Dynamic, in embedded systems. The objective is to allocate data structures from a given application, to a given set of memory banks. After proposing an integer linear program (ILP) formulation model, the chapter discusses two iterative metaheuristics, namely the long-term approach and the short-term approach, for addressing this problem. Finally, the chapter presents computational results obtained by the iterative approaches, which have been implemented in C++ and compiled with gcc 4.11 in Linux OS 10.04. The results produced by the iterative approaches are compared with the results of the ILP model and the local search method. The post hoc test shows that the ILP and the long-term approach have the same performance in terms of solution cost, but the long-term approach is better than the ILP in terms of running time.