Implementation of Farming Under PVM

Sameer S. Shende
Department of Computer and Information Science
University of Oregon

Abstract:

Keywords : dynamic load balancing, PVM, farming, heterogeneous computing, parallel and distributed processing.

• Introduction
• Load Balancing
  • Need for Dynamic Load Balancing under PVM
  • Dynamic Load Balancing
• Farming
  • Farming - Definition of terms
  • Farming - Classification
  • Goals of the farming implementation
  • Previous work in this area - Comparison
• Implementation and Design Issues
  • API definition
  • Design issues
• Results
• Future Work
  • Fault Tolerance
  • Additional farm daemons
• Conclusion
• References
• Appendix- Algorithms for implementing farming under pvm
  • int pvm_farm_init (char *farm_name);
    • Client side
    • Server side
  • int pvm_get_worker_class_id (char *farm_name, char *worker_class_name);
    • Client side
    • Server side
  • int pvm_init_worker_class(char *farm_name, char *worker_class_name);
    • Client side
    • Server side
  • Algorithm for ProcessPending() at the Server side.
  • int pvm_send_work_packet(int id);
    • Client side
    • Server side
  • int bufid = pvm_recv_work_packet(void);
    • Client side
    • Server side
  • int pvm_send_reply_packet(void);
    • Client side
    • Server side
  • int bufid = pvm_recv_reply_packet(worker_class_id);
    • Client side
    • Server side
  • int pvm_farm_terminate(char *farm_name);
    • Client side
    • Server side
  • int pvm_start_farmd(void);
    • Client side
    • Server side
  • int pvm_stop_farmd(void);
    • Client side
    • Server side
• About this document ...