Full Parallelism Research Articles

A distributed operational semantics for CCS based on condition/event systems

A new set of inference rules for the guarded version of Milner’s Calculus of Communicating Systems is proposed. They not only describe the actions agents may perform when in a given state, but also say which parts of the agents move when the global state changes. From the transition relation a particular Petri Net, namely a Condition/Event system called ΣCCS, is immediately derived. Our construction gives a semantics which is consistent with the interleaving semantics of CCS and exhibits full parallelism. The proof consists of relating the case graph of ΣCCS with the original and with the multiset (step) transition systems of the calculus.

A hardware accelerator for two-dimensional image analysis

This paper describes the architecture and operation of a new hardware accelerator called MultiRing for performing various geometrical operations on two-dimensional image space. This hardware architecture is shown to be applicable for design rule checking in VLSI layout and many image processing operations including noise suppression and contour extraction. It has both a fast execution speed and extremely high flexibility. Each row data stored in ring memory is processed in the corresponding processor in full parallelism. Each processor is simultaneously configured by the instruction decoder/controller to perform one of the 20 basic instructions each ring cycle, which gives MultiRing maximal flexibility in terms of design rule change or the instruction set enhancement. Correct functional behavior of MultiRing was confirmed by successfully running a software simulator having one-to-one structural correspondence to the MultiRing hardware.

An execution model for exploiting AND-parallelism in logic programs

This paper presents a parallel execution model for exploiting AND-parallelism in Horn Clause logic programs. The model is based upon the generator-consumer approach, and can be implemented efficiently with small run-time overhead. Other related models that have been proposed to minimize the run-time overhead are unable to exploit the full parallelism inherent in the generator-consumer approach. Furthermore, our model performs backtracking more intelligently than these models. We also present two implementation schemes to realize our model: one has a coordinator to control the activities of processes solving different literals in the same clause; and the other achieves synchronization by letting processes pass messages to each other in a distributed fashion. Trade-offs between these two schemes are then discussed.

Equations-of-motion approach to a quantum theory of large-amplitude collective motion

The equations-of-motion approach to large-amplitude collective motion is implemented both for systems of coupled bosons, also studied in a previous paper, and for systems of coupled fermions. For the fermion case, the underlying formulation is that provided by the generalized Hartree-Fock approximation (or generalized density matrix method). To obtain results valid in the semi-classical limit, as in most previous work, we compute the Wigner transform of quantum matrices in the representation in which collective coordinates are diagonal and keep only the leading contributions. Higher-order contributions can be retained, however, and, in any case, there is no ambiguity of requantization. The semi-classical limit is seen to comprise the dynamics of time-dependent Hartree-Fock theory (TDHF) and a classical canonicity condition. By utilizing a well-known parametrization of the manifold of Slater determinants in terms of classical canonical variables, we are able to derive and understand the equations of the adiabatic limit in full parallelism with the boson case. As in the previous paper, we can thus show: (i) to zero and first order in the adiabatic limit the physics is contained in Villars' equations; (ii) to second order there is consistency and no new conditions. The structure of the solution space (discussed thoroughly in the previous paper) is summarized. A discussion of associated variational principles is given. A form of the theory equivalent to self-consistent cranking is described. A method of solution is illustrated by working out several elementary examples. The relationship to previous work, especially that of Zeievinsky and Marumori and coworkers is discussed briefly. Three appendices deal respectively with the equations-of-motion method, with useful properties of Slater determinants, and with some technical details associated with the fermion equations of motion.

Unbounded hardware is equivalent to deterministic turing machines

The purpose of this paper is to investigate models of computation from a realistic viewpoint. We introduce the concept of physical computation as opposed to functional computation, and by referring to the laws of physics we study the basic criteria which a model of computation must meet in order to be realistic. With this formal apparatus, we define a very general, realistic model of planar, digital circuits, which allows for full parallelism. This model is especially well suited to describe systolic architectures. Actually, the assumption of planarity serves only practical purposes, and can be removed without altering our main results. We compare the complexity classes in this parallel model with those associated with sequential models such as the Deterministic Turing Machine (DTM) model. Our main result is that both models are space and time equivalent in the polynomial class. In particular, any circuit can be simulated in polynomial time on a DTM. One consequence is that unbounded hardware does not make NP-hardness tractable. We also address the issue of area-time tradeoffs and show that the area of a circuit can always be bounded by a polynomial function of the sequential time.

Application of microprocessor networks for the solution of diffusion equation

The availibility of low cost microprocessors, memories and ancilliary circuits has made the construction of large networks of such devices feasible where each microcomputer is dedicated to a specific task. The parallel computation of the diffusion equation is a particular application of such a multi-microcomputer network where each microcomputer simulates a given region and exchanges partial results with its neighbours. Structure of a network where each node consists of a microcomputer linked to other nodes by means of serial communication lines is described. The explicit finite difference algorithm is adapted for an 8-bit architecture most commonly available today. The algorithm is emulated on a minicomputer and the results presented. Availibility of processing power and memory at each node is significant for the simulation of non-linear problems which has been discussed. The significance of full parallelism where a microcomputer is allocated to each problem node and partial parallelism where a microcomputer simulates a number of nodes which form a region is covered.