sis 16(9): e4

Research Article

GarCoSim: A Framework for Automated Memory Management Research and Evaluation

Download889 downloads
  • @ARTICLE{10.4108/eai.14-12-2015.2262678,
        author={Konstantin Nasartschuk and Marcel Dombrowski and Tristan Basa and Mazder Rahman and Kenneth Kent and Gerhard Dueck},
        title={GarCoSim: A Framework for Automated Memory Management Research and Evaluation},
        journal={EAI Endorsed Transactions on Scalable Information Systems},
        volume={3},
        number={9},
        publisher={ACM},
        journal_a={SIS},
        year={2016},
        month={1},
        keywords={garbage collection, memory management simulation, trace file},
        doi={10.4108/eai.14-12-2015.2262678}
    }
    
  • Konstantin Nasartschuk
    Marcel Dombrowski
    Tristan Basa
    Mazder Rahman
    Kenneth Kent
    Gerhard Dueck
    Year: 2016
    GarCoSim: A Framework for Automated Memory Management Research and Evaluation
    SIS
    EAI
    DOI: 10.4108/eai.14-12-2015.2262678
Konstantin Nasartschuk1,*, Marcel Dombrowski1, Tristan Basa1, Mazder Rahman1, Kenneth Kent1, Gerhard Dueck1
  • 1: University of New Brunswick
*Contact email: kons.na@unb.ca

Abstract

Many modern programming languages rely on memory management environments that are responsible for allocation and deallocation of objects. Garbage collection phases are used in order to detect inaccessible objects on the heap so they can be deallocated. The performance of garbage collection techniques depends heavily on the environment, implementation specific parameters and the benchmark used. The contribution of this publication is an extendable memory management simulator, which aims to assist developers in memory management evaluation and research. The simulator is capable of reading operations from a trace file extracted from a virtual machine and simulating the memory management needed by the simulated mutator. The framework aims to provide an isolated experimentation and comparison platform in the field of automatic memory management. New algorithms can be added to the framework in order to compare them to established algorithms.