| B. L. Massingill. A Structured Approach to Parallel Programming. Technical Report CS-TR-98-04, Computer Science Department, California Institute of Technology, 1998. PhD thesis. |
....be done on an algorithmic level independent of the details of the specific architecture, in the same way that application development for a sequential computer can be done on an algorithmic level independent of the details of that specific architecture. Specifically, archetypes [Cha94, CMMM95, Mas98, MC96] provide a methodology and code libraries to aid in the development of correct parallel programs by exploiting common computation and communication structures. When using an archetype to write a parallel application, a programmer often needs to predict what effect modifications to an ....
....structure with a parallelization strategy gives rise to a dataflow pattern and hence a communication structure. It is this combination of computational structure, parallelization strategy, and the implied pattern of dataflow and communication that we capture as a parallel programming archetype [Mas98, MC96] For the remainder of this paper, we use the term archetype to refer to a parallel programming archetype. A key question in the development of a parallel application, especially for a multicomputer or a network of computers, is the issue of data decomposition and distribution. ....
[Article contains additional citation context not shown here]
B. L. Massingill. A Structured Approach to Parallel Programming. Technical Report CS-TR-98-04, Computer Science Department, California Institute of Technology, 1998. PhD thesis.
....except single reader single writer channels with infinite slack, if I and I 0 are two maximal interleavings of the actions of the P j s that begin in the same initial state, then I and I 0 both terminate, and in the same final state. Proof. A complete proof of this theorem appears in [20]. The technique used is to show that given interleavings I and I 0 beginning in the same state, I 0 can be permuted to match I without changing its final state. receive send receive send P1 send P0 send P0 receive P1 receive compute compute compute compute P1 compute P0 compute P1 compute P0 ....
....M implementation of the mesh archetype. Table 1 shows execution times and speedups 3 for Version C, executing on a network of Sun workstations. Figure 2 shows execution times and speedups for Version A, executing on an IBM SP. Comparable performance figures for other problem sizes are given in [20]. Execution time Speedup (seconds) Sequential 78.6 1.00 Parallel, P=1 189.0 0.41 Parallel, P=2 51.4 1.52 Parallel, P=4 25.3 3.10 Table 1. Execution times and speedups for electromagnetics code (version C) for 33 by 33 by 33 grid, 128 steps, using Fortran M on a network of Suns. Ease of use. It is ....
[Article contains additional citation context not shown here]
B. Massingill. A structured approach to parallel programming (Ph.D. thesis). Technical Report CS-TR-98-04, California Institute of Technology, 1998.
No context found.
B.L. Massingill. "A structured approach to parallel programming", Technical Report CS-TR-98-04, California Institute of Technology, 1998 (Ph.D. thesis.). ftp://ftp.cs.caltech.edu/tr/cs-tr-98-04.ps.Z.
....structure with a parallelization strategy gives rise to a dataflow pattern and hence a communication structure. It is this combination of computational structure, parallelization strategy, and the implied pattern of dataflow and communication that we capture as a parallel programming archetype [Mas98, MC96] For the remainder of the paper, we use the term archetype to refer to a parallel programming archetype. A key question in the development of a parallel application, especially for a multicomputer or a network of computers, is the issue of data decomposition and distribution. Archetypes ....
....some of the experiments described in this and other papers on archetypes [CMMM95, MC96] could be repeated. Whether an archetype based approach has similar benefits when the target architecture supports a shared memory model without performance penalties is a more difficult question. Massingill [Mas98] describes an archetype based approach to application development one of whose stages can be converted in a straightforward way to a program for a shared memory architecture, but again the experimental work focuses on the later stages of the process, whereby the original algorithm becomes a ....
B. L. Massingill. A Structured Approach to Parallel Programming. Technical Report CS-TR-98-04, Computer Science Department, California Institute of Technology, 1998. PhD thesis.
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