%0 Conference Paper %B 15th Workshop on Advances in Parallel and Distributed Computational Models, at the IEEE International Parallel & Distributed Processing Symposium %D 2013 %T Revisiting the Double Checkpointing Algorithm %A Jack Dongarra %A Thomas Herault %A Yves Robert %X Abstract—Fast checkpointing algorithms require distributed access to stable storage. This paper revisits the approach base upon double checkpointing, and compares the blocking algorithm of Zheng, Shi and Kale [1], with the non-blocking algorithm of Ni, Meneses and Kale [2] in terms of both performance and risk. We also extend the model proposed in [1], [2] to assess the impact of the overhead associated to non-blocking communications. We then provide a new peer-topeer checkpointing algorithm, called the triple checkpointing algorithm, that can work at constant memory, and achieves both higher efficiency and better risk handling than the double checkpointing algorithm. We provide performance and risk models for all the evaluated protocols, and compare them through comprehensive simulations. %B 15th Workshop on Advances in Parallel and Distributed Computational Models, at the IEEE International Parallel & Distributed Processing Symposium %C Boston, MA %8 2013-05 %G eng