The architecture and programming of a fine-grain multicomputer

[摘要] NOTE: Text or symbols not renderable in plain ASCII are indicated by [...]. Abstract is included in .pdf document.The research presented in this thesis was conducted in the context of the Mosaic C, an experimental, fine-grain multicomputer. The objective of the Mosaic experiment was to develop a concurrent-computing system with maximum performance per unit cost, while still retaining a general-purpose application span. A stipulation of the Mosaic project was that the complexity of a Mosaic node be limited by the silicon complexity available on a single VLSI chip.The two most important original results reported in the thesis are:• The design and implementation of C+-, a concurrent, object-oriented programming system.Syntactically, C+- is an extension of C++. The concurrent semantics of C+- are contained within the process concept. A C+- process is analogous to a C++ object, but it is also an autonomous computing agent, and a unit of potential concurrency. Atomic single-process updates that can be individually enabled and disabled are the execution units of the concurrent computation. The limited set of primitives that C+- provides is shown to be sufficient to express a variety of concurrent-programming problems concisely and efficiently.An important design requirement for C+- was that efficient implementations should exist on a variety of concurrent architectures, and, in particular, on the simple and inexpensive hardware of the Mosaic node. The Mosaic runtime system was written entirely in C+-.• Pipeline synchronization, a novel, generally-applicable technique for hardware synchronization.This technique is a simple, low-cost, high-bandwidth, high-reliability solution to interfaces between synchronous and asynchronous systems, or between synchronous systems operating from different clocks.The technique can sustain the full communication bandwidth and achieve an arbitrarily low, non-zero probability of synchronization failure, P[subscript f], with the price in both latency and chip area being [...].Pipeline synchronization has been successfully applied to the high-performance inter-computer communication in Mosaic node ensembles.

[发布机构] University:California Institute of Technology;Department:Engineering and Applied Science

[关键词] Computer Science