Książka w języku angielskim. The most exciting development in parallel computer architecture in recent years has been the convergence of traditionally disparate approaches namely shared-memory, message-passing, SIMD, and dataflow on a common machine structure. Parallel software has matured to the point where the popular parallel programming models are available on a wide range of machines and meaningful benchmarks exist. This maturing of the field makes it possible to undertake a quantitative and qualitative study of hardware/software interactions. This book outlines a set of issues that are critical to all of parallel architecturecommunication latency, communication bandwidth, and coordination of cooperative work (across modern designs). It describes the set of techniques available in hardware and software to address each issues and explore how the various techniques interact. Features - Provides multiprocessor systems designers with an understanding of the fundamental architectural issues and the techniques available for multiprocessor design - Provide system software designers and application developers with an understanding of this architecture to optimize software performance- Presents application case studies from computational science and engineering, computer graphics, and data mining that serve as a basis for quantitative evaluation of design trade-offs.