SIMULATION: Transactions of The Society for Modeling and Simulation International

Schlesinger, Stewart

doi: 10.1177/003754978003500401pmid: N/A

McLeod, John; McLeod, Suzette

doi: 10.1177/003754978003500407pmid: N/A

Halin, Hans J.; Bührer, Richard; Halg, Walter

doi: 10.1177/003754978003500402pmid: N/A

A new multiprocessor system is currently under con struction at ETH (Swiss Federal Institute of Technol ogy). This multiple-instruction-stream multiple-data stream (MIMD) computer exploits the advantages of low-cost large-scale integrated components. A first model of the ETH multiprocessor featuring a modular expansible architecture will consist of a PDP 11/40 and 16 LS-T-11S. After its completion in 1980 the multiprocessor will be applied mainly to the simula tion of continuous systems.Evaluations of integration schemes for MIMD systems showed the method of power series expansions to be most appropriate. Starting from a FORTRAN-like source code, the differential equations are automatically decomposed into smaller problems. The higher deriva tives of all variables are computed using analytical recursion formulas. Usually many formulas can be evaluated simultaneously, each by a master cooperating with a number of slaves, using the inherent parallel ism in many of these formulas.The time behaviour of the multiprocessor is character ized by dynamically changing hierarchies. To imple ment algorithms efficiently, several principal prob lems of parallel computing were solved in hardware rather than in software. Thus the system features provisions for instantaneous data exchange between processors, protection of data, handling of deadlocks, and performance of operating system tasks. As a result, the ETH multiprocessor simplifies the solu tion of problems from different fields.

Miner, Robin J.; Wortman, David B.; Cascio, Danis

doi: 10.1177/003754978003500403pmid: N/A

This paper describes a modular GASP IV model of a chemical plant and its use in analyzing the operation of the plant. The model represents multiple stages and products and was developed to evaluate methods for increasing throughput. The plant manufactures six products, each going through a series of steps; different products can be in production at the same time. These products differ in chemical treatment, production rate, and demand. Production must be scheduled to meet demands and to maintain minimum inventories, and it must operate within the con straints on processing space and in-process storage. This paper also includes the data-collection proce dures and the evaluation of several alternatives pro posed to increase the throughput of the plant by as much as 100%.

Chen, Wei L.

doi: 10.1177/003754978003500404pmid: N/A

Clifford, William J.; Clymer, A. Ben

doi: 10.1177/003754978003500405pmid: N/A

Kickert, Ronald N.

doi: 10.1177/003754978003500406pmid: N/A