Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....decompose the problem and to map it onto architectures. This thesis studies an important class of scientific problems, the N body problem, using hierarchical methods. We concentrate on the parallelization of the Fast Multipole Algorithm [19] one of the most efficient hierarchical methods [9]. We survey partitioning and scheduling techniques that have been employed to improve its performance on parallel computers. We fur 2 ther study and analyze the performance improvement resulting from applying Fractiling [24] a dynamic scheduling technique based on a probabilistic analysis, to ....

....step. Nearby particles are clustered together and their interaction is computed accurately whereas interactions with particles in distant clusters are computed approximately. This results in the ability to handle a large number of particles with an acceptable accuracy. A practical approximation [9] N body algorithm is the Fast Multipole Algorithm (FMA) by Greengard [19] The FMA is an O(N) hierarchical algorithm that is provably correct to any fixed accuracy. It can be applied to uniform as well as nonuniform distributions of particles. The FMA algorithm recursively constructs a tree of ....

G.E. Blelloch and J.C. Hardwick. Class Notes: Programming Parallel Algorithms. In G. Pantziou F. Makedon, P. Metaxas, editor, Notes for the DAGS'93 School on Parallel Programming, pages 118--122, June 24-29 1993.

No context found.

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

No context found.

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

No context found.

Blelloch, G. E., and Hardwick, J. C. Class notes: Programming parallel algorithms. Tech. Rep. CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, Feb. 1993.

....nested parallel functions on the Cray C90. The paper is organized as follows: Section 2 describes NESL and illustrates how nested parallelism can be applied to some simple algorithms on sparse matrices. A description of how NESL can be used for a wide variety of algorithms is given elsewhere [16]. Section 3 outlines the components of the current NESL implementation. Section 4 describes our benchmarks and Section 5 discusses the running times of the benchmarks. 3.31 1.68 3.98 2.21 Alpha C90 CM 2 CM 5 3.02 1.64 0.222 0.064 Alpha C90 CM 2 CM 5 1.45 0.059 1.57 1.06 Alpha C90 CM 2 CM 5 ....

.... in a course on parallel algorithms and has allowed students to quickly implement a wide variety of programs, including systems for speech recognition, ray tracing, volume rendering, parsing, maximumflow, singular value decomposition, mesh partitioning, pattern matching, and big number arithmetic [16]. A full implementation of NESL is available from blelloch cs.cmu.edu. The benchmark results in this paper show that it is possible to achieve good efficiency with a nested dataparallel language, across a variety of parallel machines. NESL runs within a local interactive environment that allows ....

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

....broad variety of algorithms, and on their efficiency when implemented on parallel machines. To promote the use of parallelism, Nesl supplies no serial looping constructs (although serial looping can be simulated with recursion) NESL has been used for 3 years now for teaching parallel algorithms [10], and many applications and algorithms have been written in the language [22, 4, 5] Nesl is the first data parallel language whose implementation supports nested parallelism. Nested parallelism is the ability to take a parallel function and apply it over multiple instances in parallel for ....

....the elements of a sequence, Nesl necessarily supplies the ability to apply a parallel function multiple times in parallel; we call this ability nested parallelism. For example, we could apply the parallel sequence function sum over a nested sequence: fsum(v) v in [ 2, 1] 7,3,0] 4] g; [3, 10, 4] : int] In this expression there is parallelism both within each sum, since the sequence function has a parallel implementation, and across the three instances of sum, since the apply to each construct is defined such that all instances can run in parallel. Nesl supplies a handful of functions ....

[Article contains additional citation context not shown here]

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

....be applied to some simple algorithms on sparse matrices. A description of how Nesl can be used for a wider variety of algorithms, including computing the minimum spanning tree of sparse graphs, searching for patterns in strings, and finding the convex hull of a set of points, is given elsewhere [12]. Section 3 outlines the different components of the current Nesl implementation. Section 4 describes our benchmarks and Section 5 discusses the running times of the benchmarks. 2 Nesl and Nested Parallelism Nesl is a high level language designed to allow simple and concise descriptions of ....

.... in a course on parallel algorithms and has allowed students to quickly implement a wide variety of programs, including systems for speech recognition, raytracing, volume rendering, parsing, maximum flow, singular value decomposition, mesh partitioning, pattern matching, and big number arithmetic [12]. 2 The benchmark results in this paper have shown that it is possible to get good efficiency with a nested data parallel language, across a variety of different parallel machines. Nesl runs within a local interactive environment that allows the user to execute programs remotely on any of the ....

G. E. Blelloch and J. C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, Feb. 1993.

....Our FTP site (see Section 5.1) The WWW home page contains a link to this site. ffl A mailing list used to discuss Nesl and announce new patches and releases. If you want to be added to this list send e mail to nesl request cs.cmu.edu. ffl Papers on the implementation [2] uses [4] and teaching [3] of Nesl. These can be obtained from the FTP site, or viewed directly from the WWW home page. 1.3 Conventions Within this document, interactions with the Nesl system are shown in a typewriter font, and command arguments are shown in an italic font. Enumerated choices are shown within curly ....

....read back into Nesl only the type is returned. You can use file variables in expressions just like any other variables. Here is an example: Nesl a = index(10000) a : int] Nesl sum(a) it = 49995000 : int Nesl function foo(n) take(a,n) foo = fn : int [int] Nesl foo(10) it = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] : int] File variables are stored in the temp dir specified by the configuration ( tmp by default) This means that if you switch your configuration to a new configuration with a different temp dir, the Vcode interpreter won t be able to find the variable and will give a runtime error message. ....

[Article contains additional citation context not shown here]

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

....we describe the programming language Nesl [5] which we designed based on the features, and go through several examples of how to program and analyze parallel algorithms using the language. We have been using Nesl for three years in undergraduate and graduate courses on parallel algorithms [8]. The algorithms we cover in this paper are relatively straightforward. Many more algorithms can be found through the Web version of this paper. The paper is organized as follows. Section 1 introduces and motivates the use of work and depth as complexity measures. Section 2 introduces the Nesl ....

....the current implementation reports an error. Nested parallelism is supplied in Nesl by allowing sequences to be nested and allowing parallel functions to be used in an apply to each. For example, we could apply the sum function in parallel over a nested sequence, as in fsum(a) a in [ 2,3] [8,3,9], 7] g which would return [5, 20, 7] Here there is parallelism both within each sum and across the sums. The Quicksort algorithm showed another example of nested calls the algorithm is itself used in an apply to each to invoke two recursive calls in parallel. 2.2 The Performance Model We now ....

[Article contains additional citation context not shown here]

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, School of Computer Science, Carnegie Mellon University, February 1993.

....a provably time and space efficient implementation of the parallel programming language Nesl [6] Nesl is a strongly typed call by value functional language loosely based on ML. It has been implemented on several parallel machines [8] and has been used both for teaching parallel algorithms [9, 7], and implementing various applications [17, 4, 1] The parallelism in the language is based on including a primitive sequence data type, a parallel map operation, and a small set of primitive operations on sequences. To be useful for analyzing parallel algorithms, Nesl was designed with rules for ....

Guy E. Blelloch and Jonathan C. Hardwick. Class notes: Programming parallel algorithms. Technical Report CMU-CS-93-115, Carnegie Mellon University, February 1993.

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