|
Here you could find information on different tools for multicomputers and parallel computation
COVERSConcurrent Verification and Simulation ToolCOVERS (the abbreviation stands for Concurrent Verification and Simulation tool) is a programming package for modeling and analysis of parallel and distributed real-time systems. COVERS is a graphical simulation tool with intuitive user's interface, running on highly available platform (PC under MS Windows). COVERS offers open, flexible Object-Oriented modeling methodology as a support for design and teaching in the area of discrete-event systems. The language framework includes diagrams of object structure, statecharts as a behavior description, and C++ for definition of new classes, data objects and functions. The library of Examples includes protocols (Alternating Bit protocol, a family of Multicast protocols), Distributed Election algorithm, Deadlock avoidance algorithm in Distributed Data Bases, a solution of the problem Chain of Shooters (Edward Moor's classical problem), Life game, Client-Server model, etc. COVERS model of a discrete-event system can help the user to overcome the "sequentiality" of human's way of thinking and to understand what actually happens in the complex concurrent system with the help of means for behavior visualisation. Besides, user can test the general functionality of the system and investigate the system performance. To read more about COVERS acess a Web page of the COVERS Development Group from St.Petersburg Technical University. ALTAnimating Language ToolsALT (the abbreviation stands for Animating Language Tools) is a computer tool for design and investigation cellular algorithms. ALT combines textual and graphical programming packages for representing and editing cellular algorithms, as well as simulating computational processes in cellular arrays. ALT has a multiwindow interface, a separate window being intended for the following parts of algorithm representation: a graphical window where data arrays and instruction objects are pictured, and two textual windows for writing a parallel program using a special high-level language and for writing sequential procedures with C to be executed by cells. The simulating process allows to observe computation dynamics at different levels of detail length: at program blocks, at parallel iterations, in quasiparallel mode at cell-state changes. The library of examples includes several arithmetic algorithms (addition of many numbers, multiplication in 2D and 3D arrays, vector-matrix multiplication), cellular-neural associative memory, image processing. ALT is destined for experimenting with cellular algorithms and observing distributed computation process. Of even greater importance is the possibility to improve and modify a PSA according to technical requirements, when performing architectural design of a cellular processor. ALT is running on highly available platform ( PC under MS DOS) and may be used in teaching Parallel Computing. The authors of the project will be pleased to answer any question about ALT, which may be put to O. Bandman through email address bandman@ssd.sscc.ru or to Yu. Pogudin through email address Pogudin@sbee.sunysb.edu. You could find ALT full description as well as its code at the URL http://www.ssd.sscc.ru/olga/alt/ SPARTHHigh-Accuracy Computations with Dynamically Changed Length of OperandsSPARTH-MIMD is a portable Super-precision Parallel ARiTHmetic package for MIMD computers based on Message Passing Interface for communications. This tool has been designed to support High-Accuracy Computations with Dynamically Changed Length of Operands on a broad variety of parallel computers available today. The package was written with C programming language and developed so that it should be used with usual C programs. It has all needed arithmetic and arithmetic-logical operations, operations for capacity and resource menagement, service and special functions. A broad scope of data transformation and input/output procedures allows not only exclude the data representation errors but also use mixed (for example, double precision and SPARTH) computations. The structure of the package lets it to be used on both sequential as well as parallel machines. To port the tool on another platform communication procedures need to be re-written only. The package is now installed on Parsytec PowerXPlorer system, iPSC-860 from Intel, MVS-100 Intel860 based multicomputer from Kvant, T800 based multitransputer system. To use the tool in sequential mode one may have a copy for IBM PC, Silicon Graphics or Sun computers. C++ and Fortran (for Microsoft Fortran on IBM PC) are also available.
The package includes the Reference Manual with examples allowing to
demonstrate how to use it.
|