Enabling Large Scale Irregular Computations of Sparse Data

Maria Jesus Garzaran, University of Illinois at Urbana-Champaign

Usage Details

As the amount of data increases, the processing of large data sets is becoming challenging, specially, ?when the data representation encountered by the large-scale data-intensive applications use some form?of sparse array storage format.??In this proposal, we will evaluate the performance of data-intensive irregular sparse computations when ?running on the novel and highly scalable architecture of Blue Waters. Two applications will be evaluated. ?The first one is a hybrid parallel linear solver that solves large sparse linear systems from a variety of domains. ?The second one measures the Betweeness Centrality (BC) of a variety of big networks and graphs.