Showing papers in "Future Generation Computer Systems in 1990"
••
TL;DR: An optimized technique to perform effective broadcasting operations on networks belonging to the WK-recursive class is described, one prototype of which has been realized at the Hybrid Computing Research Center.
31 citations
••
TL;DR: The architecture of DISC, a system for parallel software development, is described, designed for programming computer systems having several autonomous units, not memory-sharing, and linked by means of a communication network.
16 citations
••
TL;DR: In this article, it is argued that neural networks (cell assemblies) whose connectivity involves strong feedback and which consequently can sustain patterns of high spiking activity (reverberations), are plausible descriptions of cognitive function in certain cortical regions.
12 citations
••
TL;DR: Initial studies into the smart memory mechanism are presented with the goal of describing its potential and stimulating further work.
12 citations
••
TL;DR: These two methods for job scheduling in parallel computer systems have the advantages of lower real time scheduling overhead, less context switching, less communication delay and providing deadlock prevention.
6 citations
••
TL;DR: The problems of processor self-scheduling schemes in a preemptive environment are described and a solution for resolving these problems is presented.
6 citations
••
TL;DR: A novel translation technique is introduced that parameterizes the translation process according to the attributes of the remote architectures, operating systems, and software applications that maintain the data.
4 citations
••
TL;DR: A new framework for parallel execution of logic programs is presented, including the OR-forest description and the abstract process model, which can exploit AND-OR-parallelisms in both deterministic and nondeterministic programs.
4 citations
••
TL;DR: Through compiling and running some typical benchmarks in SES-PIM, it is concluded that the compiling approach can, in most cases, exploit as much AND-parallelism as the dynamic approach does under “producer-consumer” scheme, and needs less dynamic overhead while exploiting more AND- parallelism than DeGroot's approach.
4 citations
••
TL;DR: A partial differential equation is derived from the microscopic dynamics given by the master equation for the position- and time-dependent correlations between the system state and the stored patterns, which can be used to study networks with finite range connections, the behaviour of domain boundaries and information transport.
3 citations
••
TL;DR: A tool which, by monitoring the communications between processes allocated to separate processing elements, is able to provide indications on the effectiveness of the partitioning and mapping actions carried out is described.
••
TL;DR: The C-Oriented tool ZIP/E++ is developed to verify the authors' ideas on what characteristics such tools should have and is a flexible tool for integrating different kinds of inference engines and knowledge representations into a single system.
••
TL;DR: It is proven that for a given group of benchmarks the speed of the prototype is over 20 times higher than a PROLOG system running on an Intel 86/310.
••
TL;DR: An integrated database architecture is presented, where the two sub-databases are tightly combined at the clause level, and therefore can be managed uniformly, and a uniform implementation of Prolog database operations is presented.
••
TL;DR: Methods of how clauses are represented in compiler-based Prolog systems are presented, i.e. source-copying based, source-assertion based and decompilation based.
••
TL;DR: A model combining dataflow with control flow is presented, and a practical architecture SDS-1 according to DFGC is described, which exploits two levels of parallelism.
••
TL;DR: Close scrutiny of the test results shows that basic architectural differences may explain the performance characteristics of the Ardent Titan-2 and the Stellar GS1000 in these benchmarks.
••
TL;DR: Using the concepts of the Abstract SUPRENUM Machine and some programming environment tools these applications can be parallelized rather easily and straightforward.
••
TL;DR: Schemes for managing Prolog databases and algorithms for implementing database operation built-ins in the framework of the WAM-Plus are presented.
••
TL;DR: The structure of the attribute set is discussed and an attribute theory for machine learning is presented and an algorithm AL is given for learning by analogy based on attribute theory.
••
TL;DR: Four examples are given to show how ALT can be used to implement many concepts of software engineering and artificial intelligence, they are intuitionistic logic, Peano arithmetic, approximate reasoning and program transformation.
••
TL;DR: It is shown through simulations that the simple static scheduling policy does reasonably well and the performance cannot be significantly improved by the dynamic policy which is more complex and incurs higher overheads.
••
TL;DR: A framework for a quotient structure model of hierarchy is presented and a new statistical heuristic search technique is introduced as an application of this model.
••
TL;DR: LAIOS (Lattice for Artificial Intelligence Oriented System) is a project of a multiprocessor architecture oriented to artificial intelligence applications that permits one to understand the rules and the parameters that govern the parallel execution of PROLOG programs on this lattice and using this model.