A Bridging Model for Parallel Computation

Introduction

• The success of Von Neumann Model of sequential computation is attributable to the fact that it is an efficient bridge between software and hardware.
• A similar Analogous bridge is required for parallel computation if that is to become widely used.
  • A major purpose of such a model would be simply to act as a standard on which people can agree
  • The model however has to satisfy some stringent quantitative requirements, exactly as does the Von Neumann model

Bulk Synchronous Parallel (BSP) Model

• It is intended not as a hardware nor a programming model, but something in between
• A major feature of BSP -> Provides the programmer with the option to avoid the burdens of managing memory, assigning communications and performing low level synchronization while maintaining optimal efficiency ( within constant factors)
  • Provided that the programmer writes programs with sufficient parallel slackness
  • This means that, programs are written for v virtual processors, to be run on p parallel processors, such that \(v\ = \ p\ log\ p\)
  • The slack here is exploited by the compiler to schedule and pipeline the computation and communication efficiently.