MPICH-V: Toward a Scalable Fault Tolerant MPI for Volatile Nodes

By George Bosilca, A. Bouteiller, Franck Cappello, Samir Djilali, Gilles Fedak, C. Germain, Thomas Hérault, Pierre Lemarinier, O. Lodygensky, Frédéric Magniette and 2 more. International Conference on Software Composition, 2002

355 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Hardware and Architecture

View Paper: https://www.semanticscholar.org/paper/d5045481d61b8ad1cfa45076997cc188017b80f6

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Hardware and Architecture

Citations in the corpus

Open MPI: Goals, Concept, and Design of a Next Generation MPI Implementation (2004, background): “If the application is to survive a process failure without having to restart from the beginning, it either must regularly write checkpoint files (and restart the application from the last consistent checkpoint [1, 8]) or the application itself must be able to adaptively handle process failures durin”
DMTCP: Transparent checkpointing for cluster computations and the desktop (2007, background): “1) User-level MPI libraries for checkpointing [4], [5], [12], [14], [15], [32], [34], [36], [37]: works for distributed processes, but only if they communicate exclusively through MPI (Message Passing Interface).”
Piccolo: Building Fast, Distributed Programs with Partitioned Tables (2010, background): “There has been continuous research on developing experimental features for MPI, such as optimization of collective operations [3], fault-tolerance via machine virtualization [34] and the use of hybrid checkpoint and logging for recovery [10].”
Addressing failures in exascale computing (2014, background): “aimed at providing a fault-tolerant MPI have been going on for over a decade [44, 18], and several prototype implementations of fault-tolerant versions of MPI exist.”
Proactive fault tolerance for HPC with Xen virtualization (2007, methodology): “A number of systems have been developed that combine FT with the message passing implementing MPI, ranging from automatic methods (checkpoint-based or logbased) [32, 29, 6] to non-automated approaches [3, 14].”
Toward Exascale Resilience (2009, background): “(Bosilca et al. 2002), local area multicomputer message-passing interface (LAM-MPI) (Indiana) (LAM/MPI, 2009), MPVAPICH (Ohio) (MVAPICH, 2009), Los Alamos message passing interface (LA-MPI)”