andysomogyi / mdcore Goto Github PK

mdcore is small, self-contained Open-Source platform-independent library for
Molecular Dynamics simulations. This project aims not to provide yet another
full-featured simulation package, but rather to provide the core
computational infrastructure for parallel MD in such a way that it can
be easily integrated into existing simulation packages or be used
directly for simple simulations. The library also provides a test bed
for the design and development of new algorithms for new architectures 
such as GPUs [1], the Cell/BE [2], or the Intel MIC, and to investigate
different strategies for hybrid shared/distributed-memory parallelism,
asynchronous communication and load balancing for multi-core architectures
on low-bandwidth/high-latency commodity networks.

mdcore currently incorporates the following functionality:

    * Task-based shared-memory parallelism for multi-core computers,
    * Hybrid shared/distributed memory parallelism for clusters of
      multi-core computers,
    * Support for multiple CUDA-capable GPUs,
    * Support for the Cell/BE,
    * Support for arbitrary potential functions.
    
mdcore was originally written by Pedro Gonnet, and is currently maintained by
Andy Somogyi. 

The source code and documentation are currently hosted on GitHub. To
download the latest version, explore the tutorials, or read the documentation,
please visit the site

    https://andysomogyi.github.io/mdcore/