C. Foster, "Content Addressable Parallel Processors," Van Nostrand-Reinhold, New York, 1976.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....some of the limitations of the system, and to bring the technology up to date. 2. Low Level Processor Design Tradeoffs in the First Generation IUA Our approach to the design of the low level processor was based on our experience with content addressable, or associative, parallel processors [6] for real time applications. Our analysis of low level vision algorithms showed that the majority would best be served by a meshconnected array, augmented with the features of an associative processor (i.e. global broadcast with local partial matches, activity control with global override, and ....

Foster, C. C., Content Addressable Parallel Processors, New York: Van Nostrand Reinhold, 1976.

....very small processor. The interconnection network provides communications among the processing elements, and between the set of cells and the control processor. Control Processor Interconnection Network cell cell cell cell cell cell cell cell cell Foster describes early data parallel machines [2]. The Massively Parallel Processor [7] and the Connection Machine [4] are recent data parallel machines that have been fabricated commercially. This paper uses Apsa [6] a simpler architecture designed specifically for symbolic and functional languages. The main differences among these ....

Foster, Caxton C. Content Addressable Parallel Processors, New York: Van Nostrand Reinhold Co. (1976).

....tag bits) or selecting a single responder for further processing. For example, the controller 2 may execute the instruction, How many cells have variable GREEN set to TRUE or All elements with BROWN set to TRUE set SKY to FALSE. The basic associative operations are summarized below (after [3, 13]) 1. Global broadcast local compare activity control 2. Some None of responders from array to controller 3. Count of responders from array to controller 4. Select a single responder In multiassociative processing, we add the concept of a set, which we define as a collection of elements that ....

C.C. Foster (1976): Content Addressable Parallel Processors, Van Nostrand Reinhold Co. New York.

....have a mechanism to summarize array information through a global OR mechanism. This allows the host to query the array to nd out, e.g. whether any PE has found a result. Foster and Weems have demonstrated the bene ts to associative processing of a feedback mechanism that counts responding PEs [7, 37]. This latter feature was available in the Clip 4 [8] and the CAAPP [37] among others. Array Instruction Issue Building a system where the PEs run at maximum possible speeds requires very careful attention to instruction issue and distribution. This problem is complicated if, as is generally ....

Foster, C. C. Content Addressable Parallel Processors. Van Nostrand Reinhold Co., New York, NY, 1986.

.... network processing [Higuchi et al. 1991] and maintaining hypothesis blackboards in a multi agent knowledge based system for speech understanding [Asanovi c and Chapman, 1988] Associative processors promise significant improvements in cost performance for these data parallel AI applications [Foster, 1976, Lea, 1977, Kohonen, 1980] SPACE is an associative processor architecture designed to allow experimentation with such applications. A large SPACE array has been built as part of the PADMAVATI project [Guichard Jary, 1990] The core of PADMAVATI is a MIMD transputer array, where each processor ....

....bit columns as well as specific words. This ability to selectively write individual bit columns, together with the ability to write multiple selected words in parallel, distinguishes a content addressable parallel processor (CAPP) such as SPACE from a less powerful content addressable memory (CAM) [Foster, 1976]. There are two write instructions: wal writes to all selected words, wfi writes only to the first selected word (if any) The flags of the written words are updated to the value of NF. Read instructions There is a single read instruction rfi which returns the 36b value of the first selected ....

[Article contains additional citation context not shown here]

Foster, C. C. Content Addressable Parallel Processors. Computer Science Series. Van Nostrand Reinhold, 1976.

....single bit register across all PEs. This capability is sufficient to tell whether a data dependent computation has completed. A more complex summary is the return of the count of those register values. This additional information is used for efficient implementation of associative algorithms [58, 62, 144, 72] and for making data dependent algorithmic decisions [70] 34 3.2.4 InterPE Communication All the processors considered have hardware support for inter PE communication among nearest neighbors in a two dimensional mesh configuration. We also consider some processors that have more ....

Foster, C. C. Content Addressable Parallel Processors. Van Nostrand Reinhold Co., New York, NY, 1986.

....at Amherst Amherst, Massachusetts 01003 NetAd : herbordt cs.umass.EDU Abstract Computation on and among data sets mapped to irregular, non uniform, aggregates of processing elements (PEs) is a very important, but largely ignored, problem in parallel vision processing. Associative processing [11] is an effective means of applying parallel processing to these computations [33] but is often restricted to operating on one data set at a time. What we propose is an additional level of parallelism we call multi associativity as a framework for performing associative computation on these data ....

....the Content Addressable Array Parallel Processor (CAAPP) 35] to simulate efficiently within aggregates of PEs simultaneously the associative algorithms typically supported in hardware at the array level. Some of the results are: the efficient application of existing associative algorithms (e.g. [10, 11]) to arbitrary aggregates of PEs in parallel, and the development of new multi associative algorithms, among them parallel prefix and convex hull. The multi associative framework also extends the associative paradigm by allowing operation on and among aggregates themselves, operations not defined ....

[Article contains additional citation context not shown here]

C.C. Foster (1976): Content Addressable Parallel Processors, Van Nostrand Reinhold Co. New York.

....In particular, distributed associative memories have been described by a number of workers (Willshaw 1969, Kanerver 1988, Kohonen 1984) However, no one has extended the concept to distributed computation. Work on associative processors has restricted its self to non distributed processing (Foster 1976). More conventional parallel processor work is constrained to a single process per processor per time instant. Pre cursors to the PDC have appeared in Austin (1992) and will appear in Austin and Jackson (1992) where the use of the ADAM network is shown to be applicable to reasoning with imprecise ....

C Foster, Content Addressable Parallel Processors, 1976, Van Nostrand Reinhold Company.

.... being used during the reduction, the complexity of step 2 can range from O(log N) in the general case, down to O(1) when executing the CountSelectedResponders operation with appropriate hardware support [28] Similarly, step 1 can also have complexity from O(1) using hardware described in [8] to O(log N) if only a global OR circuit is available. In any case, the algorithm is efficient if and only if jRj is small. We have seen, however, that the number of regions in an oversegmentation can be huge: in the case where regions are relatively small, jRj = O(N ) The idea behind hybrid ....

C.C. Foster, Content Addressable Parallel Processors, New York: Van Nostrand Reinhold Co., 1976.

....are graphics (e.g. the HP VISUALIZE PxFl) compression [20] neural nets [15] image processing [3, 7, 16] DSP [5] MMX, and chess (the Deep Blue coprocessors) ffl Avoiding the memory wall by building PEs onto memory cores. Although integrating processing and memory is an old idea (see e.g. [6]) the latest projects have access to current technology and are yielding impressive results. See e.g. 9, 7] ffl Developments in processing technology. Once an entire SIMD array is built on a single chip, signal propagation problems will be simplified by several orders of magnitude. Since it is ....

Foster, C. C. Content Addressable Parallel Processors. Van Nostrand Reinhold Co., New York, NY, 1986.

....superstructures upon this workhorse. What stands in the way of the advancement of specialized hardware In our opinion, there are three major kinds of reasons hampering the development in this field. 1. Psychological reasons. It is worth mentioning the explanation given one day by Caxton Foster [7] of the reason hindering broad application of associative processors: the same reason that seems to inhibit all new computer architectures. This is the conservatism in its original sense of computer programmers. They look at the design, find it difficult to understand because of its very ....

....types VVP, VoO. 5.5 Associative Array Processors Different Associative Array Processors (AAPs) were developed to implement non numerical problems (database machines, information systems, etc. Extensive literature is devoted to the architecture and applications of AAPs (see, for instance, [7,19]) The core of each AAP is some version of contentaddressed memory (CAM) A classical distributed CAM represents a strongly specialized processor implementing the basic procedure equality search . As has been shown by various authors, the AAP can realize different algorithms and problems. ....

Foster C.C. Content Addressable Parallel Processors, New York, Van Nostrand Reinhold, 1976.

No context found.

C. Foster, "Content Addressable Parallel Processors," Van Nostrand-Reinhold, New York, 1976.

No context found.

C. Foster, "Content Addressable Parallel Processors," Van Nostrand-Reinhold, New York, 1976.

No context found.

C. C. Foster, "Content Addressable Parallel Processors", Van Nostrand Reinhold Co., 1976, chs. 2 & 5.

No context found.

C. Foster. Content Addressable Parallel Processors. Van Nostrand Reinhold Co, New York, 1976.

No context found.

Foster, C.: Content Addressable Parallel Processors. Van Nostrand Reinhold Co, New York (1976)

No context found.

C.C.Foster, "Content Addressable Parallel Processors, " Van Nostrand Reinhold, New York, (1976).

No context found.

C. C. Foster, Content Addressable Parallel Processors, Van Nostrand Reinhold Co., New York, (1976).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC