scientific article; zbMATH DE number 3924088
asynchronous processesdeadlockfreedistributed decentralized controlintercomputer linkslivelockfreemultiplexing of linksmultiprogram functioning of computersoperating systemparallel computer systemsparallel programssemaphoressynchronizing primitivevirtual computersvirtual linksVLSI
Research exposition (monographs, survey articles) pertaining to computer science (68-02) Theory of operating systems (68N25) Models and methods for concurrent and distributed computing (process algebras, bisimulation, transition nets, etc.) (68Q85) Applications of graph theory to circuits and networks (94C15)
On the basis of physical constraints it is shown that VLSI are the best means to construct parallel computer systems with programmable structure and distributed decentralized control. Based on the analysis of requirements to commutators used in such systems, classes of graphs are singled out which assign optimal, by the fault-tolerance property, intercomputer links. Established is the expediency of applying graphs with degree of vertices in the range 4-8. Considered are the means of organizing multiprogram functioning of computers and multiplexing of links. Singled out is the synchronizing primitive based on the extended concept of semaphores and most suitable to control creation, destruction and interaction of asynchronous processes running in computer systems with programmable structure. Presented is the approach to decentralized hierarchical deadlockfree and livelockfree distribution of resources among parallel programs coming from peripherial devices distributed among the computers. As units of resources selected to perform parallel programs of users and the operating system, subsystems are introduced consisting of virtual computers and of virtual links connecting them. Considered are the ways to organize intercomputer exchanges and the way to embed parallel programs into subsystems with an arbitrary graph of intercomputer links. Given are parallel programs illustrating the parallel programming style. The author presented requirements to computer systems. The realization of them is necessary to gain the acceleration of parallel computations which is directly proportional to the number of computers used. In general, the book is an attempt of complex approach to solving problems arising in creation of decentralized control of computations and in organizing effective realization of parallel programs.
- High level of synthesis parallel problem-oriented TS -- Coordinated digital computing systems
- scientific article; zbMATH DE number 3846828 (Why is no real title available?)
- scientific article; zbMATH DE number 3858379 (Why is no real title available?)
- On mapping graphs of parallel programs onto graphs of distributed computer systems by recurrent neural networks
- Parallel programming
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3698264)