M. J. Flynn. Some Computer Organisations and their Effectiveness. IEEE Transactions on Computers, C-21(9), 1972.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....properties after construction. Programs should be architecture independent to allow reuse. The model should support cost measures to guide the design process and should have guaranteed performance over a useful variety of architectures. Flynn has classified typical parallel data structures in [32], where SIMD is an abbreviation for Single Instruction, Multiple Data and MIMD for Multiple Instruction, Multiple Data. Depending on what information is explicit in a model they distinguish six levels, i.e. 1. nothing explicit 2. parallelism explicit 3. parallelism and decomposition explicit ....

M. J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9):948--960, September 1972.

....degree of parallelism of an APM program, maximized over all admissible input files of size n. Note that the , parallelism of APMs derives from the implicit parallelism of identical operations and not from any parallelism of instructions. Thus it is a prototypical SIMD machine (see e.g. Flynn [7]) in its basic mode of operation. In the model we assume that there is no priori bound on the maximum length of the vectors that can be processed in one instruction, to facilitate the study of vectortzed algorithms without further conceptual complication. If a maximum length L is in effect, then ....

....will be assumed. A SIMDAG (Goldschlager [9] consists of a CPU, a set of parallel processing units (PPU 0, PPU1, with local memory (y locations) and a global random access memory (x locations) The PPU s carry their own index in a signature register. The SIMDAG is an SIMD machine (Flynn [7]) The CPU executes a SIMDAG program serially, and either applies an instruction directly to global memory (indirect addressing allowed) or broadcasts it to an initial segment of the PPU s for parallel execution in the local memories or on global memory (with indirect addressing only via ....

Flynn, M.J., Some computer organisations and their effectiveness, IEEETrans. Comput' C21(1972) 948-960. - 29-

....we will give most emphasis to the hardware architecture. For a description of machines that can be considered to be classified as high performance one is referred to [20] and, for more recently available systems, 19] 2. 1 The main architectural classes Since many years the taxonomy of Flynn [6] has proven to be useful for the classification of high performance computers. This classification is based on the way of manipulating of instruction and data streams and comprises four main architectural classes. We will first briefly sketch these classes and afterwards fill in some details when ....

M.J. Flynn, Some computer organisations and their effectiveness, IEEE Trans. on Comp., Vol. C-21, 9, (1972) 948--960.

....computing, one first identifies those parts of the computation which can be performed simultaneously, and organises the computation over a collection of communicating processors. Parallel hardware architectures can be characterised under a number of different headings. Flynn s classification [8] is based on the instruction and data streams in the machine: are identical instructions broadcast to each processor (Single Instructionstream Multiple Data stream SIMD) or is each processor independently programmable (Multiple Instruction stream Multiple Data stream MIMD) Do the processors ....

M. J. FLYNN, Some computer organisations and their effectiveness, IEEE Trans. Comput. C-21 (1972) pp. 948-960.

....Caches can dramatically reduce the average time it takes for the CPU to obtain the information it requires. 1.1. 2 Shared memory parallel architectures Flynn defined two types of parallel architecture: Single Instruction, Multiple Data (SIMD) and Multiple Instruction, Multiple Data (MIMD) [12]. For the MIMD model, an additional distinction can be made based on the programming 1.1. Background 17 model. With distributed memory, the memory space is partitioned between processors, meaning that the programmer (or the compiler or operating system) has to explicitly arrange for data transfer ....

Michael J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9):948--960, September 1972.

....structures [32] 30] These attempts using the inverted file model, showed considerable performance, even under the above limitations and the limited power parallel environment they used. Concerning the type of parallelism, we can distinguish between the SIMD and MIMD environments (due to Flyn [11]) Most parallel text retrieval systems [30] 31] 32] have been implemented on the Connection Machine, which is a classical type of SIMD architecture using pipelined vector processors. However the last few years have seen increasing interest in retrieval systems for use in MIMD environments, ....

M. Flynn, "Some Computer Organisations and their Effectiveness", IEEE Transactions on Computers, C--21, pp. 948--960, 1978.

....are given. Finally, in Chapter 5, we present our conclusions and suggestions for future work. 7 Chapter 2 Background Information In this chapter, we present background information for this report. 2. 1 Classification Multiprocessors are classified into SIMD and MIMD by Flynn s taxonomy [29] by data and instruction streams. But further classification is required to differing modern MIMD multiprocessors which have shown different significant characteristics. Considering the physical organisation of memory and the organisation of the address spaces on individual processors, MIMD can be ....

M.J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9), 1972.

....could be enhanced, and the paper concludes with a summary of our main findings. 2 THE TRANSPUTER 2. 1 Transputer hardware and software There have been several attempts to classify the many ways in which parallelism may be implemented in a computer system (see, for example, References [11 13]) in this paper, we NEAREST NEIGHBOUR SEARCHING USING TRANSPUTER NETWORKS 187 are concerned with what are generally referred to as multiple instruction stream, multiple data stream,orMIMD computers [11] which are characterized by multiple processors capable of independent operation [14 16] The ....

.... in which parallelism may be implemented in a computer system (see, for example, References [11 13] in this paper, we NEAREST NEIGHBOUR SEARCHING USING TRANSPUTER NETWORKS 187 are concerned with what are generally referred to as multiple instruction stream, multiple data stream,orMIMD computers [11], which are characterized by multiple processors capable of independent operation [14 16] The most widely applied sub categorization of MIMD computers depends on the degree to which the processor memories are coupled. In tightly coupled systems, all of the processors share a global memory ....

M. J. Flynn, `Some Computer Organisations and their Effectiveness', IEEE Transactions on Computers, C-21, 948--960 (1972).

....of development shows no signs of slacking, with CHAPTER 2. PARALLEL COMPUTING 25 manufacturers continuing to appear and disappear every year. A number of attempts have been made to develop a formal taxonomy of parallel architectures (see, for example, 25] and [27] though only that of Flynn [15] has gained universal acceptance. Flynn distinguishes two primary classes of parallel computer, Single Instruction Multiple Data (SIMD) machines and Multiple Instruction Multiple Data (MIMD) machines. As the names imply, SIMD machines apply the same instruction concurrently to many data items, ....

Flynn, M.J., (1972) Some computer organisations and their effectiveness, IEEE Trans. Comput., vol. C-21, pp. 948--960. BIBLIOGRAPHY 184

....structures ( 15] 16] These attempts using the inverted file model, showed considerable performance, even under the above limitations and the limited power parallel environment they used. Concerning the type of parallelism, we can distinguish between the SIMD and MIMD environments (due to Flyn, [18]) Most parallel text retrieval systems ( 10] 15] 16] have been implemented on the Connection Machine, which is a classical type of SIMD architecture using pipelined vector processors. However the last few years have seen increasing interest for parallel retrieval systems under MIMD ....

M. Flynn, "Some Computer Organisations and their Effectiveness", IEEE Transactions on Computers, C--21, pp. 948--960.

....architectures which depart radically from von Neumann model were designed and constructed. Although there are different methods used to classify computers, it seems there is no single characterisation which fits all designs. Perhaps the most widely used classification is that described by Flynn [Flyn72] which uses the relationship of program instructions to program data. Flynn categorised computers into four types based on the instruction data relationship. These are SISD (Single Instruction, Single Data) SIMD (Single Instruction, Multiple Data) MISD (Multiple Instruction, Single Data) of ....

MJ Flynn, Some Computer Organisations and their Effectiveness. IEEE Transactions on Computers, 21, pp. 948-960, September 1972.

....and peripherals. The goal of a computer architect is to lay out these functional units so as to maximise throughput at all levels, avoiding any bottlenecks in the architecture. This has led to several classifications of computer organisation. The best known classification is the one due to Flynn [67] which identifies two main classes of multiprocessors: Single Instruction Multiple Data(SIMD) and Multiple Instruction Multiple Data (MIMD) Many commercial multiprocessors from both classes are currently available. Over the recent years, MIMD multiprocessors with significantly differing ....

....practical shared memory paradigms and the various consistency models is presented. 2.1 Classifications and taxonomies of multiprocessor Architectures 6 2. 1 Classifications and taxonomies of multiprocessor Architectures The most widely used taxonomy of computer architectures is due to Flynn [67]. Flynn s criterion for classification is based on the two basic activities of a computer system, i.e. instruction execution and data access. This leads to the Single Instruction, Multiple Data stream (SIMD) and Multiple Instruction Multiple data stream (MIMD) classes of multiprocessors. SIMD ....

M. J. Flynn. Some Computer Organisations and their Effectiveness. IEEE Transactions on Computers, C-21(9), 1972.

....of development shows no signs of slacking, with CHAPTER 2. PARALLEL COMPUTING 25 manufacturers continuing to appear and disappear every year. A number of attempts have been made to develop a formal taxonomy of parallel architectures (see, for example, 25] and [27] though only that of Flynn [15] has gained universal acceptance. Flynn distinguishes two primary classes of parallel computer, Single Instruction Multiple Data (SIMD) machines and Multiple Instruction Multiple Data (MIMD) machines. As the names imply, SIMD machines apply the same instruction concurrently to many data items, ....

Flynn, M.J., (1972) Some computer organisations and their effectiveness, IEEE Trans. Comput., vol. C-21, pp. 948--960. BIBLIOGRAPHY 184

....such as communication overheads and initialization costs. To understand the factors better and to get an overview of parallel architectures one has to CHAPTER 1. INTRODUCTION 20 look at the taxonomy of parallel architectures. The first attempt to classify computer architectures was made by Flynn [9, 10]. 1.4 Flynn s Hardware Taxonomy Flynn has attempted to classify computer architectures, serial as well as parallel, into four basic categories. These categories are SISD (single instruction single data) SIMD (single instruction multiple data) MISD (multiple instruction single data) and MIMD ....

....has attempted to classify computer architectures, serial as well as parallel, into four basic categories. These categories are SISD (single instruction single data) SIMD (single instruction multiple data) MISD (multiple instruction single data) and MIMD (multiple instruction multiple data) In [10] Flynn has justified the inclusion of special cases in one of the four categories mentioned above. For example, pipelined architectures have been classified as time multiplexed versions of processor arrays and are hence included in the SIMD class. A more detailed description of these categories is ....

[Article contains additional citation context not shown here]

M. J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers Vol. 21, No. 9, pages 48--60, September 1972.

....on each classifier or can be parallelised by using some standard data parallel algorithms. Also genetic operators, such as crossover and mutation, can be applied independently on distinct genotypes, or couples of genotypes. Data parallelism is a natural programming paradigm for SIMD (Flynn, 1972) architectures, but it can also be used on MIMD machines. On the other hand, a MIMD architecture allows many simultaneously active flows of control operating on different data sets, thus it is also suitable for control parallelism. Control parallelism can be exploited in a CS by overlapping ....

Flynn, M. J. (1972). Some computer organisations and their effectiveness. IEEE Transaction on Computers, C-21(9):948--960.

....No Home Location Fixed Home Location DDM, KSR1 (COMA) DASH, Alewife etc. Figure 1: Different forms of shared memory abstractions coherence in DSM systems. A better solution to the problem of inconsistent terminology would be to develop a more objective taxonomy similar to the one invented by Flynn [35] to classify (multi) processor architectures. We propose one such taxonomy as follows. Just as Flynn based his classification on the two fundamental activities in a computer system, we use the address space and the physical memory organisation as the basic entities that describe a memory system ....

M. J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9), 1972.

No context found.

M. J. Flynn. Some Computer Organisations and their Effectiveness. IEEE Transactions on Computers, C-21(9), 1972.

No context found.

M. J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9):948--960, September 1972.

No context found.

M. J. Flynn. Some computer organisations and their effectiveness. IEEE Transactions on Computers, C-21(9):948--960, September 1972.

No context found.

M. Flynn, "Some Computer Organisations and their Effectiveness", IEEE Transactions on Computers, C--21, pp. 948--960.

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