This is an MPI version of the GH-RDF3x.

Changes over GH-RDF3x: - Required compiler GCC47 and OPENMPI-1.4.3 (preferred) - Includes two new folders tools/rdf3xvpload and tools/rdf3xvpquery - Makefiles have been modified in the corresponding directories to accomodate the new folders


  • Command usage: bin/rdf3xvpload sample_input.nt
  • This code only accepts RDF graphs which are in .nt (ntriples) format
  • Name stands for rdf3x-vertex-partition-load. Each MPI task is responsible for creating its own partition and finally a db-filei.e., the number of partitions is equal to the number of MPI tasks specified.
  • The ouput will be n partitioned .nt files along with their db files
  • For debugging purpose, a boolean global varaible runMPI is set to false by default. This results in creating a single partition when you run the above command.


  • Command usage: mpirun -n numb_of_nodes bin/rdf3xvpquery inputFile.nt
  • You should use the same input file name which was used for rdf3xvpload.
  • Running this prompts you for a query. Sample can be found in the file 'sample_queries' in the base directory, for which the solutions are available in 'sample_solutions'
  • Similar to rdf3xvpload, rdf3xvpquery loads each patitioned database in a separate MPI task.
  • Task-0 broadcasts the input query and each task runs the query independently


GH-RDF3X is a more complete and modified version of the original RDF-3X engine. Web site:

RDF-3X was created by: Thomas Neumann. Web site: (c) 2008 RDF-3X was modified by: Hancel Gonzalez and Giuseppe De Simone (Advisor: Maria Esther Vidal). (c) 2013