The STE||AR Group

License: Boost Software License (BSL1.0)

Web Page: https://github.com/STEllAR-GROUP/hpx/wiki/GSoC-2014-Project-Ideas

Mailing List: hpx-users@stellar.cct.lsu.edu

STE||AR (pronounced as stellar) stands for “Systems Technologies, Emergent Parallelism, and Algorithms Research”. We are a group of faculty, researchers and students working at the Center of Computation and Technology (CCT - www.cct.lsu.edu) at Louisiana State University (LSU - www.lsu.edu). Everything we do is centered around HPX - A general purpose C++ runtime system for parallel and distributed applications of any scale. We use HPX for a broad range of scientific applications, helping scientists and developers to write code which scales better and shows better performance if compared to more conventional programming models such as MPI or OpenMP.

Projects

• Create an HPX backend for Thrust Thrust is the library for parallel algorithm and Data structure. Thrust provide the STL like API for GPU programming that helps programmer to develop the high performance application which exploit parallelism with minimum programmer effort. Thrust use the CUDA for NVDIA GPU and TBB or OpenMP for multi-core CPU as backend but there is no backend implemented in HPX for thrust.These project is intended to implement the backend for thrust in HPX.
• Implement a sparse matrix container and an unstructured grid for LibGeoDeco I'm planing to extend the functionality of the LibGeoDecompi[1] to handle irregular grids and graphs. Therefor I want to implement a new unstructured grid class derived from the BaseGrid. To save the adjacent information I want to introduce a sparse Matrix container based on the SELL-C-σ format for efficient access and memory usage on SIMD units. [1] http://www.libgeodecomp.org