Charm++ is a C++-based parallel programming system, founded on the migratable-objects programming model, and supported by a novel and powerful adaptive runtime system. It supports both irregular as well as regular applications, and can be used to specify task-parallelism as well as data parallelism in a single application. It automates dynamic load balancing for task-parallel as well as data-parallel applications, via separate suites of load-balancing strategies. Via its message-driven execution model, it supports automatic latency tolerance, modularity and parallel composition. Charm++ also supports automatic checkpoint/restart, as well as fault tolerance based on distributed checkpoints.
Charm++ is a production-quality parallel programming system used by multiple applications in science and engineering on supercomputers as well as smaller clusters around the world. Currently the parallel platforms supported by Charm++ are the IBM BlueGene/Q and OpenPOWER systems, Cray XE, XK, and XC systems, Omni-Path and Infiniband clusters, single workstations and networks of workstations (including x86 (running Linux, Windows, MacOS)), etc. The communication protocols and infrastructures supported by Charm++ are UDP, MPI, OFI, Infiniband, uGNI, and PAMI. Charm++ programs can run without changing the source on all these platforms. Charm++ programs can also interoperate with MPI programs (§23). Please see the Installation and Usage section for details about installing, compiling and running Charm++ programs (§A).
- 1.1 Programming Model
- 1.2 Execution Model
- 1.3 Proxies and the charm interface file
- 1.4 Machine Model