
	Where results make sense

Topic: Parallel algorithm

	Spartanburg SC \| GoUpstate.com \| Spartanburg Herald-Journal
		In computer science, a parallel algorithm, as opposed to a traditional serial algorithm, is one which can be executed a piece at a time on many different processing devices, and then put back together again at the end to get the correct result.
		Parallel algorithms are valuable because it is faster to perform large computing tasks via a parallel algorithm than it is via a serial (non-parallel) algorithm, because of the way modern processors work.
		A subtype of parallel algorithms, distributed algorithms are algorithms designed to work in cluster computing and distributed computing environments, where additional concerns beyond the scope of "classical" parallel algorithms need to be addressed.
	www.goupstate.com /apps/pbcs.dll/section?category=NEWS&template=wiki&text=Parallel_algorithm (536 words)

	Parallel algorithm Summary
		Parallel algorithms running on many machines thus are required--by partitioning the large problem among many processors (such as by allocating large chunks of the three-dimensional segment space to them for individual processing), problems such as weather forecasting may be solved.
		In computer science, a parallel algorithm, as opposed to a traditional serial algorithm, is one which can be executed a piece at a time on many different processing devices, and then put back together again at the end to get the correct result.
		Parallel algorithms are valuable because it is faster to perform large computing tasks via a parallel algorithm than it is via a serial (non-parallel) algorithm, because of the way modern processors work.
	www.bookrags.com /Parallel_algorithm (1312 words)

	Parallel Architectures
		The inherent problem for parallel algorithm designers is how to parallelize an algorithm allowing for a given number of processors while minimizing the size or the depth of the algorithm.
		The size of the algorithm relates to the number of operations which must be performed for a desired outcome.
		The parallel prefix is performed on the new combined evens and then the evens are combined with the next higher odds at output.
	carbon.cudenver.edu /~galaghba/parallel.html (675 words)

	PSTSWM Algorithm Comparision II
		Parallel Algorithm Comparison II A very large number of parallel algorithms is supported in PSTSWM, and it is time consuming to examine all possibilities in order to determine the optimal algorithms.
		In a distributed algorithm the original decomposition of the domain is retained, and communication is performed to allow the processors to cooperate in the calculation of a transform.
		A parallel algorithm for PSTSWM consists of pairing a parallel algorithm for the Fourier transform with a parallel algorithm for the Legendre transform.
	www.csm.ornl.gov /~worley/studies/algorithmsII.html (929 words)

	A Parallel Algorithm
		We now consider the next stage of the algorithm in which the third superblock, for which the data distribution is shown explicitly, is factored.
		The parallel implementation closely follows the sequential implementation presented in Section 3.
		Pseudocode for the parallel version is given in Figure 8.
	www.netlib.org /utk/papers/outofcore/node12.html (450 words)

	A Parallel Algorithm for DNA Alignment
		The algorithm described here allows for large genomes to be aligned by making efficient use of memory resources and by utilizing parallelization techniques.
		This is done by first sorting all of the matches based on their position in input sequence s via a sample sort algorithm [12] and consolidating the results onto a single node N. Node N finds the longest increasing subsequence (LIS) of these matches based on their position in input sequence t [5].
		Unfortunately, no parallel algorithm is known for the LIS problem but for highly homologous sequences it can be computed in linear time with regards to number of matches to be considered.
	acm.org /crossroads/xrds9-3/alignment.html (2970 words)

	A Parallel Algorithm for DNA Alignment
		The algorithm described here allows for large genomes to be aligned by making efficient use of memory resources and by utilizing parallelization techniques.
		This is done by first sorting all of the matches based on their position in input sequence s via a sample sort algorithm [12] and consolidating the results onto a single node N. Node N finds the longest increasing subsequence (LIS) of these matches based on their position in input sequence t [5].
		Unfortunately, no parallel algorithm is known for the LIS problem but for highly homologous sequences it can be computed in linear time with regards to number of matches to be considered.
	www.acm.org /crossroads/xrds9-3/alignment.html (2970 words)

	Parallel optimization
		Our algorithms allow for the most efficient usage of existing computational resources because the number of processors actively involved in solving the problem is independent of the problem dimension.
		Our algorithms make it realistic to formulate and solve the optimization problems even when many hours are required to the response values for one combination of design variables (for example, 3D CFD codes).
		The main difference between the developed parallel optimization algorithm and the basic IOSO algorithm is in the information received by the data analysis and moving strategy unit.
	www.iosotech.com /parallel.htm (581 words)

	Probability and Algorithms
		A parallel algorithm that is efficient (or optimal) when run using a certain number of processors win remain an efficient (or optimal) parallel algorithm when implemented on a smaller number of processors while running proportionately slower.
		Hence it is of interest to construct efficient parallel algorithms that run very fast, regardless of the number of processors available.
		The last is a subroutine used in many efficient parallel algorithms to ensure that all of the processors perform about the same amount of work.
	www.nap.edu /openbook.php?record_id=2026&page=154 (490 words)

	A Library of Parallel Algorithms
		The algorithms are implemented in the parallel programming language NESL and developed by the Scandal project.
		The parallel algorithms are based on the idea of contracting the graph.
		Spanning-tree algorithms are similar to those for connected-components, except that spanning-tree algorithms need to keep track of which edges are used for contraction and they do not need to expand the graph back out.
	www-2.cs.cmu.edu /~scandal/nesl/algorithms.html (497 words)

	Parallel Implementation of the Filtered Back Projection Algorithm for Tomographic Imaging
		Parallel implementation of CT algorithms became important when it became necessary to reconstruct 3D images in real time [23].
		Since the serial algorithm carries out the FFT computations row-wise on each of the matrices P(n) and since the FFT operations on one row are independent from the other rows, it was decided to parallelize by distributing an equal number of rows to each processor.
		Since the FFT algorithm works best for vectors which are integer powers of 2, zero padding has been done to keep the vector length of each row as an integer multiple of 2.
	www.sv.vt.edu /xray_ct/parallel/Parallel_CT.html (6430 words)

	[No title]
		The foundation of parallel programming is concurrency: the condition of a problem or a system in which two or more tasks are active simultaneously.
		This problem is a task-level parallel problem since you are using the multiple tasks to define the parallelism.
		Once you have a parallel algorithm, you need to translate the algorithm into source code and run it on a parallel computer.
	www3.intel.com /cd/ids/developer/asmo-na/eng/219575.htm?page=2 (1001 words)

	Programming on Parallel Computing Architectures
		Because the cost of processors is decreasing, parallel processing, the ability to execute several computations at once to solve a problem, has become an increasingly viable solution for success in this endeavor [2].
		The results of this algorithm are such that each PE will have the average value of its surrounding PEs on each iteration, thereby spreading the value of the heated location across the plate.
		The two parallel computing architectures presented here show how parallel computing performs, and also shows considerations that must be kept in mind when applying algorithms to parallel systems.
	www.public.asu.edu /~vidar/parallel/parallel.html (3641 words)

	Citations: HIFI: From Parallel Algorithm to Fixed-Size VLSI Processor Array - Held, Dewilde, Deprettere, Wielage ...
		Usually, in existing design systems the algorithm is represented as a system of ane recurrence equations de ned in di erent computation spaces.
		Held, P. Dewilde, Ed Deprettere, and P. Wielage, "HIFI: From Parallel Algorithm to Fixed-Size VLSI Processor Array", Application-Driven Architecture Synthesis, Francky Catthoor and Lars Svensson, Ed's, 1993, pp.
		Node c represents the condition(also known as fork node [3] and the triangle is a dummy node, representing the join node corresponding to node c.
	citeseer.ist.psu.edu /context/322827/0 (1225 words)

	1.4 Parallel Algorithm Examples
		We do not concern ourselves here with the process by which these algorithms are derived or with their efficiency; these issues are discussed in Chapters 2 and 3, respectively.
		The goal is simply to introduce parallel algorithms and their description in terms of tasks and channels.
		Algorithm 1.1 explores a search tree looking for nodes that correspond to ``solutions.'' A parallel algorithm for this problem can be structured as follows.
	www-unix.mcs.anl.gov /dbpp/text/node10.html (1319 words)

	Parallel Algorithm Design
		The parallel algorithm for a given problem attempts to divide it into sub-problems which can then be solved concurrently on the different processors of a parallel computer.
		In order to design a parallel solution to this problem, it must first be decomposed into smaller tasks which can be executed simultaneously.
		Parallel algorithm design is an interesting and challenging area of computer science which requires a combination of creative and analytical skills.
	www.dcs.ed.ac.uk /home/stg/pub/P/par_alg.html (510 words)

	PSTSWM Algorithm Comparision I
		In this study we compare the performance of the different parallel algorithms within a subset for each platform and message-passing library, determining both which parallel algorithms are optimal (and when), and the sensitivity of performance to the choice of the parallel algorithm.
		The "larger" dimension is used to examine the parallel Fourier or Legendre transform options, while a fixed parallel algorithm is used for the "smaller" dimension.
		The default parallel algorithm for the other dimension uses a communication protocol that is generally one of the worst performers.
	www.csm.ornl.gov /~worley/studies/algorithmsI.html (1514 words)

	Dr. Dobb's \| Programming Paradigms \| July 22, 2001
		Any parallel algorithm can be sequentialized, with a total time of the order of magnitude of the sum of the times taken by all the processors m the parallel version.
		Simulated annealing, the Metropolis algorithm, neural networks, Boltzmann machines, and Parallel Distributed Processing are among the heuristic-based techniques that have attracted a legion of algorists as motley as the father of the H-bomb and the codiscoverer of the genetic code.
		The parallel version of the algorithm is able to break the rules and achieve superlinearity, the authors say, because solutions are distributed nonuniformly on the average.
	www.ddj.com /184407988;jsessionid=Q4DLLZPYH1QLOQSNDLQSKH0CJUNN2JVN?_requestid=774229 (2406 words)

	Fine-grained Parallel Genetic Algorithms in Charm++
		Genetic algorithms [2,4] are a robust optimization tool that can be used to solve a wide range of difficult problems efficiently and accurately.
		Fine-grained genetic algorithms divide the population into small subpopulations containing only one or a couple of solutions which are connected in a grid topology (usually, 2D grids are used).
		In our implementation of the fine-grained genetic algorithm, the population of the candidate solutions is mapped to a 2D grid where each position of the grid may either contain a particular solution or be empty.
	www.acm.org /crossroads/xrds8-3/fineGrained.html (3854 words)

	Factors That Limit Speedup
		A Parallel Algorithm is an algorithm for the execution of a program which involves the running of two or more processes on two or more processors simultaneously.
		The latter is the fairest way to compare parallel algorithms but it is unrealistic in practice, since most people do not have access to the fastest serial machines, making it impossible to make a claim about speedup.
		Since most parallel programs contain a certain amount of sequential code, a possible conclusion of Amdahls Law is that it is not cost effective to build systems with large numbers of processors because sufficient speedup will never be produced.
	www.cs.cf.ac.uk /Parallel/Year2/section7.html (1378 words)

	A Library of Parallel Algorithms
		The algorithms are implemented in the parallel programming language NESL and developed by the Scandal project.
		The parallel algorithms are based on the idea of contracting the graph.
		Spanning-tree algorithms are similar to those for connected-components, except that spanning-tree algorithms need to keep track of which edges are used for contraction and they do not need to expand the graph back out.
	www.cs.cmu.edu /~scandal/nesl/algorithms.html (497 words)

	Reevaluating Amdahl's Law and Gustafson's Law
		It requires the serial algorithm to retain its structure such that the same number of instructions are processed by both the serial and the parallel implementations for the same input.
		Often the parallel implementation is directly crafted from the corresponding serial implementation of the same algorithm.
		A sequential algorithm is non-structure persistent (NSP) if there exists at least one parallel implementation of the same algorithm, at least one input, that the parallel implementation requires less total number of calculation steps (including those in parallel) than the total pure sequential steps.
	www.cis.temple.edu /~shi/docs/amdahl/amdahl.html (2909 words)

	parallel_processors
		Given a parallel algorithm that uses p processors (p can be a function of the size n of the problem) and terminates in time T
		The efficiency of the algorithm measures the fraction of the time that a typcial processor is usefully employed.
		The Parallel Random Access Machine or PRAM is a theoretical model of parallel computation in which an arbitrary but finite number of processors can access any value in an arbitrarily large shared memory in a single time step.
	campus.murraystate.edu /academic/faculty/bob.pilgrim/405/parallel_processors.html (1699 words)

	Citations: A fast and simple randomized parallel algorithm for the maximal independent set problem - Alon, Babai, Itai ...
		However, it was soon realized that O(1) wise independent binary random variable constructions were inadequate to derandomize many parallel algorithms; n random variables that are roughly O(log n) wise independent seemed to....
		In all of these algorithms a reduction in randomness is traded for suboptimal performance.
		Alon, L. Babai and A. Itai, A fast and simple randomized parallel algorithm for the maximal independent set problem, Journal of Algorithms 7 (1986), pp.
	citeseer.ist.psu.edu /context/30310/0 (1698 words)

	The Parallel Genetic Algorithm
		This section describes the distributed memory version of a parallel genetic algorithm for the training of neural networks as it has been implemented for a transputer network of 16 T800's.
		The genetic algorithm consists of two main parts, the evaluation of the fitness and the fitness based selection and generation of a child population.
		The genetic algorithm involves only communications between master and slaves, which should be reflected in an appropriate topology of the transputer network.
	tph.tuwien.ac.at /~oemer/doc/neurogen/node25.html (410 words)

	MPIKAIA - Parallel Genetic Algorithm
		A genetic algorithm based fitting procedure quite naturally divides into two basic functions: evaluating the modeling fitness-function, and applying the genetic operators to each generation once the fitnesses have been calculated.
		Moreover, it requires minimal transfer of information, since all that the user-supplied function requires is the n-dimensional floating-point array of parameters defining a single instance of the model, and all it needs to return is the floating-point value corresponding to the model's fitness.
		This dual-functionality requires a front end program (mpi_pikaia.f) to determine whether a given instance is the master or the slave, to call the appropriate code as a subroutine, and to terminate all of the jobs when
	www.hao.ucar.edu /Public/about/Staff/travis/mpikaia (1229 words)

	Method for a parallel-sort algorithm hardware implementation
		Some fast sorting algorithms, such as quick sort and shell sort, are in some way derived from bubble sorting.
		The algorithm maximizes the parallelism of sorting, using a minimum of hardware resources, while providing implementation flexibility.
		The method regroups and parallelizes all compare-and-swap operations required for a bubble sort to achieve sorting in the shortest time.
	www.priorartdatabase.com /IPCOM/000141641 (287 words)

	The Parallel Algorithm
		The CPU and memory requirements of the algorithm grow exponentially with the number of dimensions.
		The main disadvantage of the parallelization inside the simplex is that the cost of the communication between two processes in PVM is very high.
		The implemented parallel program is based on a master-slave structure, where the master program distributes the phase space to smaller pieces (domains) and the slaves figure out the equation system in these domains.
	www.fsz.bme.hu /~szebi/pvm/node4.html (559 words)

Try your search on: Qwika (all wikis)