2: Hello World in GraphLab

To use GraphLab,

  #include <graphlab.hpp>

All of GraphLab lives in the graphlab namespace. You may use

  using namespace graphlab;

if you wish, but we recommend against it.

Your main function should begin and end with:

  int main(int argc, char** argv) {
    graphlab::mpi_tools::init(argc, argv);
    graphlab::distributed_control dc;
    
    ... main body ...
    graphlab::mpi_tools::finalize();
  }

dc is the distributed communication layer which is needed by a number of the core GraphLab objects, whether you are running distributed or not.

Placing the following code in my_first_app.cpp running "make" in the debug/ release/ build folders will create a program which when executed, will print "Hello World!".

  #include <graphlab.hpp>
  int main(int argc, char** argv) {
    graphlab::mpi_tools::init(argc, argv);
    graphlab::distributed_control dc;
 
    dc.cout() << "Hello World!\n";

    graphlab::mpi_tools::finalize();
  }

dc.cout() provides a wrapper around standard std::cout, but wraps it in a way that when used in a distributed environment, only one copy will print, even though all machines execute it. To try that run the following:

  mpiexec -n 4 ./my_first_app

This should run 4 instances of "my_first_app" all on the local machine. However, only one "Hello World!" will be printed.

Note:

The distributed_control object is the core RPC implementation and has many other capabilities. See GraphLab RPC for details.

In the next section, we will see how to define a distributed graph.