an extensible parallel graph library for graph analytics

 The Combinatorial BLAS (CombBLAS) is an extensible distributed-memory
 parallel graph library offering a small but powerful set of linear
 algebra primitives specifically targeting graph analytics.
 .
   - The Combinatorial BLAS development influences the Graph BLAS
     standardization process.
   - It achieves scalability via its two dimensional distribution and
     coarse-grained parallelism.
   - CombBLAS powers HipMCL, a highly-scalable parallel implementation of
     the Markov Cluster Algorithm (MCL).
   - Operations among sparse matrices and vectors use arbitrary user
     defined semirings.
 .
 This package provides the combblas shared library.