STAPL Components
Related Projects:  STAPL: Standard Template Adaptive Parallel Library    STAPL Applications    STAPL Graph Library    pAlgorithms    pContainers    pViews  
Current Contributors: Francisco Coral, Lawrence Rauchwerger, Nancy Amato
Project Alumni: Adam Fidel, Alireza Majidi, Timmie Smith, Nathan Thomas, Harshvardhan, Joshua Wright, Mustafa Abdul Jabbar, Sam Ade Jacobs, Ioannis Papadopoulos, Gabriel Tanase, Francisco Arzu, Antal A. Buss, Nicolas Castet, Dielli Hoxha, Tarun Kumar Jain, Vincent Marsy, Steven Saunders, Shishir Sharma, Xiabing Xu, Daniel Latypov

Supported By: Supported by NSF and the DOE

The Standard Template Adaptive Parallel Library (STAPL) comprises several sections: Run-time System, Paragraph, Skeletons Framework, pContainers, pViews, and pAlgorithms.

STAPL's run-time system (RTS) provides to the Developer and Specialist the following facilities:

  • Communication primitives, based on Adaptive Remote Method Invocation (ARMI);
  • Executor of pRange's tasks that enforces tasks dependencies;
  • User definable scheduler for the tasks of pRanges;
  • Performance Monitor for adaptiveness and for user feedback.

    ARMI communication library hides all the details about the underlying platform, by being implemented over the lower level communication facilities, such as MPI, OpenMP, Pthreads, etc. The communication interface supports a well defined consistency model to allow the developers to design algorithms in a uniform and portable way.

    PARAGRAPH is the STAPL data flow engine which allows parallelism to be expressed explicitly using data flow graphs (a.k.a. task graphs).

    The STAPL Skeleton Framework is an interface for algorithm development which helps developers to only focus on defining their computation in terms of skeletons. Each skeleton is translated to a parametric data flow graph and is expanded upon the presence of input data. The data flow representation of skeletons allows programs to run on distributed and shared memory systems.

    A pContainer is a distributed data structures that have interfaces similar to the (sequential) C++ standard library (stl). Its data is partitioned and distributed across the machine, but the User is offered a shared object view. The pContainer distribution across the machine can be user specified or automatically selected by STAPL. STAPL provides a large number of basic data parallel structures (e.g., pArray, pList, pVector, pGraph, pMap, pSet).

    pViews are the STAPL equivalent of STL iterators in the sense that they provide a generic mechanism to access the data of the pContainers. pViews emphasize processing data ranges over accessing single items. Each pView may be partitioned into subviews hierarchically and this allow to adjust the degree of parallelism to the application needs and nested parallelsm.

    A pAlgorithm is the parallel equivalent of an STL algorithm. A pAlgorithm is written in terms of pViews operations. The hierarchical structure of input pViews and the algorithm access pattern decide the degree of parallelism available for computation. A pAlgorithm can modify input pViews for optimized data access and/or easier algorithm specification.

    The STAPL Graph Library (SGL) is a distributed-memory high-performance parallel graph processing framework written in C++ using STAPL. In addition to a graph data structure, SGL includes a collection of efficient parallel graph algorithms.




  • STAPL Graph Library


    The STAPL Graph Library (SGL) is a distributed-memory high-performance parallel graph processing framework written in C++ using STAPL. In addition to a graph data structure, SGL includes a collection of efficient parallel graph algorithms.

    pAlgorithms


    Written in terms of Views operations, a pAlgorithm is the parallel equivalent of an STL algorithm. It can modify input Views for optimized data access and/or easier algorithm specification. The hierarchical structure of input Views and algorithm access pattern decide the degree of parallelism for computation.

    pContainers


    A pContainer is the parallel equivalent of an STL container. Its data is partitioned and distributed across the machine, but the User is offered a shared object view. The pContainer distribution across the machine can be user specified or automatically selected by STAPL.

    pViews


    pViews, or only Views, are the STAPL equivalent of STL iterators in the sense that they provide a generic mechanism to access the data of the pContainers. Views emphasize processing data ranges and provide iterators to access single pContainer elements.