Boots – shoes – and leggings
Patent
1995-03-22
1997-09-23
Shah, Alpesh M.
Boots, shoes, and leggings
395650, 3642292, 3642303, 3642383, 364DIG1, G06F 1300
Patent
active
056714300
DESCRIPTION:
BRIEF SUMMARY
FIELD OF THE INVENTION
The invention relates to an intelligent communications method, as well as to means for performing the method in multiprocessor systems, e.g. for use in electronic data processing.
BACKGROUND OF THE INVENTION
Nowadays multiprocessor systems are used in increasing numbers of large and small computers and for very varied applications. Thus, e.g. a personal computer is provided with several specialized processors for the operation of the display screen, keyboard, mass memory, etc., i.e. each personal computer is already a multiprocessor system.
Multiprocessor systems are naturally not only used in personal computers, but also in computing-intensive applications, for scientific calculations such as meteorology and fluid mechanics or for digital image processing. Such applications can be very well performed by parallel operating processors, i.e. by means of several or even better a plurality of processors, which are interconnected by means of a communications system. It can therefore be assumed that in future multiprocessor systems will play an important part.
Standard supercomputers such as e.g. the CRAY Y-MP or the NEC SX-3 have computing capacities of several GFLOPS (Billion Floating Point Operations per Second). A single modern processor, such as the intel Processor i860, the PowerPC or the Motorola DSP96002, or also the DEC Alpha only have a computing capacity of up to and above 100 MFLOPS (Million Floating Point Operations per Second). The GFLOP threshold is only reached by a synergy-giving coupling of a larger number of individual processors to form a parallel computer by means of a communications system. This parallel electronic data processing is subject to difficulties in communication and programming. The nature of the information switching (communication) consequently has a great influence on the efficiency and construction of parallel computers.
The communication difficulties are a general problem of and are inherent in multiprocessor systems. In order to obtain a high computing power the maximum number of processors must cooperate. In principle, a distinction can be made between four different architectures.
a) Shared memory, in which all the processors access a single, common memory. The synchronization of the processors takes place with locked and unlocked instructions. The bottleneck in such a system is the common memory. Therefore this architecture is only suitable for multiprocessor systems with a relatively small number of processors.
b) Distributed memory, in which each processor has an individual memory, which can also be accessed by the other processors. Each access consists of a request (address) and the actual data value. In the case of intensive communication with many accesses of several processors the corresponding communications system is rapidly overloaded.
c) Client/server, where a master (client) distributes the tasks and data to several slaves (servers). Such an architecture can only be used if the individual tasks and data have no or only very few interactions with one another, which only occurs with a few applications.
d) Message passing, where the data produced in a processor are supplied by means of a message to all processors interested in said specific data. The number of messages supplied in this communications system can grow in proportion to the square. Therefore this architecture is also unsuitable for tasks with intensive communication and many processors.
These communications systems are characterized by a high management expenditure and low data throughput. As can be gathered from the above remarks, central significance is consequently attached to the transmission protocol. The speed with which the communication can become a bottleneck is e.g. shown by the symmetrical multiprocessor system constituted by the Intel Pentium, in which up to four processors can communicate with one another. The multiprocessor Interconnect (MPI) Communications System used has nine different cycles by means of which the individual processors can converse in accor
REFERENCES:
patent: 4631667 (1986-12-01), Zulian et al.
patent: 4873626 (1989-10-01), Gifford
patent: 5321825 (1994-06-01), Song
patent: 5390336 (1995-02-01), Hillis
patent: 5423008 (1995-06-01), Young et al.
patent: 5517656 (1996-05-01), Shi
Hutchinson, Norm et al, "The Flooding Sink--A New Approach to Local Area Networking", Computer Networks and ISDN Systems, vol. 11, No. 1, Jan. 1986 (Elseive3r Science Publishers B.V., Amsterdam, Netherlansds, pp. 1-14.
"Parallel Data Transfer Operations", IBM Technical Disclosure Bulletin, vol. 34, No. 10B, Mar. 1992, pp. 245-246.
"Data Structures and Algorithms for Collective Communication Over a Dynamically Determined Set of Processors in Distributed Memory Parallel Computers", IBM Technical Disclosure Bulletin, vol. 35, No. 3, Aug. 1992, pp. 465-470.
Farley Walter C.
Shah Alpesh M.
LandOfFree
Parallel data processing system with communication apparatus con does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Parallel data processing system with communication apparatus con, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Parallel data processing system with communication apparatus con will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1942397