Supercoprocessors (SCPs), dedicated VLSI systems for solving a small class of problems, are proposed as attractive alternative to present-day supercomputers. Two SCPs for sorting, one based on a generalized odd-even t...
详细信息
ISBN:
(纸本)0818608021
Supercoprocessors (SCPs), dedicated VLSI systems for solving a small class of problems, are proposed as attractive alternative to present-day supercomputers. Two SCPs for sorting, one based on a generalized odd-even transposition sort and the other implementing a systolic tree sort, are described. Notions of problem-size impedence, elasticity, and limited bandwidth constraint are introduced, and a lower bound for parallel sorting under these assumptions is derived.
A metalanguage for denotational semantics is given a logical interpretation: types are interpreted as propositional theories;terms (in an extended typed lambda -calculus, denoting elements of types) are embedded in an...
详细信息
ISBN:
(纸本)0818607939
A metalanguage for denotational semantics is given a logical interpretation: types are interpreted as propositional theories;terms (in an extended typed lambda -calculus, denoting elements of types) are embedded in an endogenous program logic, which characterizes their behavior in terms of which properties they satisfy;terms (in an extension of the algebraic metalanguage of Cartesian closed categories, denoting morphism between types) are embedded in an exogenous logic, which characterizes their behavior as predicate transformers or modal operators. This interpretation is related to the standard domain-theoretic one by means of the machinery of Stone duality, and an exact correspondence is obtained. Some applications to logics for specific computational situations (e. g. , concurrency) are mentioned.
The optical matrix-vector multiplier described uses a Hughes liquid crystal light valve as the active element, the residue arithmetic number system and a holographic table lookup. It is designed with the premise that ...
详细信息
ISBN:
(纸本)0936659661
The optical matrix-vector multiplier described uses a Hughes liquid crystal light valve as the active element, the residue arithmetic number system and a holographic table lookup. It is designed with the premise that only one light valve response time is to be used to optimize throughput and that there are a plethora of image elements available on the light valve. We describe the mapping approach to residue arithmetic and how it is implemented in principle and in actuality including the role of a holographic lookup table. Results of operation are shown.
It is proved for parallel sorting that six three-sorters are optimal for merging three sorted sequences of equal size three. A depth-first search algorithm is used.
It is proved for parallel sorting that six three-sorters are optimal for merging three sorted sequences of equal size three. A depth-first search algorithm is used.
The problem of adding timing constraints to the generalized input-output construct of C. A. R. Hoare's CSP (1978) is discussed. Correctness criteria are identified and an algorithm is presented and shown to be cor...
详细信息
ISBN:
(纸本)0818608013
The problem of adding timing constraints to the generalized input-output construct of C. A. R. Hoare's CSP (1978) is discussed. Correctness criteria are identified and an algorithm is presented and shown to be correct. The solution is also compared with existing solutions to the problem of generalized input-output without timing constraints, i. e. , when the deadline is infinite.
An area-efficient, high-speed shuffle bus suitable for VLSI interprocessor communication and sorting networks is presented. The bus provides a data-swapping rate between adjacent registers of up to 200 MHz (in 1. 25- ...
详细信息
ISBN:
(纸本)0818608021
An area-efficient, high-speed shuffle bus suitable for VLSI interprocessor communication and sorting networks is presented. The bus provides a data-swapping rate between adjacent registers of up to 200 MHz (in 1. 25- mu m CMOS) and can be constructed as a bidirectional pipelined bus for VLSI processor arrays. It is simple and modular in structure and can be operated in both synchronous and asynchronous modes. It may also be expanded to accommodate any number of nodes with an arbitary number of bits in each word. A bit-serial sorting algorithm (called converge-sort) with average time performance of O(N) is derived. When it is organized as a mesh-connected array, the shuffle bus improves the implementation feasibility of many existing VLSI sorting algorithms.
A programming environment for robotics based on a threaded code architecture is presented. The environment is supported by a multiprocessing operating system and includes a programming language, an interactive interpr...
详细信息
A programming environment for robotics based on a threaded code architecture is presented. The environment is supported by a multiprocessing operating system and includes a programming language, an interactive interpreter, a debugger, and a multiwindow manager. It is proven that the threaded code organization on which the system is based is beneficial in obtaining a good integration of the programming tools.
The author considers how to prepare user documentation in a system development environment. He suggests that the key word that characterizes the system development environment is change, and that it is this facet of t...
详细信息
The author considers how to prepare user documentation in a system development environment. He suggests that the key word that characterizes the system development environment is change, and that it is this facet of the environment that challenges the writer. He includes a description of the system development environment, which he views as fundamental to the document development cycle.
Summary form only given, as follows. The notions of indistinguishability and lack of discrepancy are captured by maximal fixed points. Results in concurrent processes and operational semantics are discussed.
ISBN:
(纸本)0818607939
Summary form only given, as follows. The notions of indistinguishability and lack of discrepancy are captured by maximal fixed points. Results in concurrent processes and operational semantics are discussed.
An empirical study examined the effectiveness of four different versions of a self-instruction manual for a word processing system: a Skeletal version that explicitly states only the essential information, an Inferent...
详细信息
ISBN:
(纸本)0897912136
An empirical study examined the effectiveness of four different versions of a self-instruction manual for a word processing system: a Skeletal version that explicitly states only the essential information, an Inferential version that has the user infer some of the essential information, a Rehearsal version that is like the Skeletal manual, but adds opportunities to rehearse the explicitly stated information, and a Lengthy version that adds nonessential explanatory and descriptive information to the Skeletal version. The best learning performance was obtained with the inferential approach, particularly for more realistic tasks.
暂无评论