Libraries and Algorithms: Standard Libraries for Parallel Programming
Our objective is to implement a rich set of efficient, re-usable parallel libraries, thereby reducing the time needed to develop parallel applications and encourage modular programming. The libraries are designed to be versatile enough to accommodate the varying demands of the programmer, and still highly efficient in that the programmer need not pay for the features that he is not using. Attempts will be made to make the run-time system take decisions at run time and choose the most appropriate library for the particular problem at hand.
  • Reductions
  • Communication
  • TeMPO
  • Load Balancing Algorithms
  • PVM
  • Simple Messaging
  • Master/Slave
  • PE Sync
  • Branch & Bound
  • Heap, Set, Extensible FIFO
People
Papers/Talks
99-06
1999
[Paper]
Branch and Bound Based Load Balancing for Parallel Applications [LNCS 1999]
95-13
1995
[Paper]
A Parallel Adaptive Fast Multipole algorithm for N-body problems [ICPP 1995]
95-11
2005
[Paper]
Efficient, Language-Based Checkpointing for Massively Parallel Programs [PPL Technical Report 2005]
95-07
1995
[Paper]
Modularity, Reuse, and Efficiency with Message-Driven Libraries [SIAM Conference on Parallel Processing for Scientific Computing 1995]