A Parallel-Design Distributed-Implementation (Pddi): General-Purpose Computer (Classic Reprint) - Hardcover
Synopsis
Explore how a parallel-design, distributed-implementation computer model makes algorithms faster on many processors.
This book introduces a general-purpose parallel computer that blends a synchronous shared-memory design with a distributed execution model. It shows how algorithms written for one model can be translated into another, and why this helps designers build scalable, efficient machines.
The discussion centers on the synchronous distributed computer (SDC) and its way of organizing computation with many processing elements connected in a fixed pattern. It explains how common memory access and specialized networks—like sorting and merging networks—enable fast data movement and coordination. The text also covers practical techniques, such as pipelining, to improve performance when simulating traditional parallel models on a distributed architecture. Throughout, the focus remains on design space, implementation space, and how to translate between them.- Learn the key ideas behind a Fetch-and-* operation and how it works with shared memory.
- See how sorting and merging networks are used to implement parallel algorithms efficiently.
- Understand how a guided simulation reduces processor and memory requirements while keeping correctness.
- Discover how pipelining and dynamic resource assignment help manage communication bottlenecks.
"synopsis" may belong to another edition of this title.
