Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 1.27 KB

README.md

File metadata and controls

20 lines (14 loc) · 1.27 KB
Raw

Analytical Applications

As examples, we provide six applications built with libgrape-lite following the guide of LDBC Graph Analytics Benchmark. They are, single-source shortest path(SSSP), weakly connected component(WCC), PageRank, local clustering coefficient(LCC), community detection of label propagation(CDLP), and breadth first search(BFS). Each application comes with a few variants, from which advanced users can learn how to customize an application in different ways (message handing, local parallelization, ...).

To run a specific analytical application, users may use command like this:

# run sssp with 4 workers in local.
mpirun -n 4 ./run_app --vfile ../dataset/p2p-31.v --efile ../dataset/p2p-31.e --application sssp --sssp_source 0 --out_prefix ./output_sssp --directed

# or run weakly connected component with 4 workers on a cluster.
# HOSTFILE list hosts on which to launch MPI processes. 
mpirun -n 4 -hostfile HOSTFILE ./run_app --application=wcc --vfile ../dataset/p2p-31.v --efile ../dataset/p2p-31.e --out_prefix ./output_wcc

# see more flags info.
./run_app --help

Users may also want to run a benchmarking suite with the ldbc-driver. Please refers to here for more details.