Independent partitioning of algorithms with uniform dependencies

Shang, Weijia; Fortes, J.A.B.

doi:10.1109/12.123395

Cited by 50 publications

(21 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Shang and Fortes [22] introduce the concept of pseudo-dependences to characterize dependence relations that do not consider both the boundary conditions of the iteration space and the direction of the dependence vectors. Two computations v 1 Most existing techniques for exploiting parallelism within loops are based on pseudo-dependences.…”

Section: B Motivationmentioning

confidence: 99%

“…The program model considered in this paper is the uniform dependence loop nests [1,22,24] whose dependence pattern is the same for each iteration of loops. A loop iteration is considered as the basic scheduling unit and is called a computation.…”

Section: A Terminologymentioning

confidence: 99%

“…A great deal of research has gone into the development of various techniques for exploiting parallelism within nested loops [3-11, 13-18, 20-27]. Some well-known techniques are wavefront execution [11], hyperplane partitioning [5], time transformation [21], affine scheduling [3], partitioning vector approach [22], minimum distance [14], and loop labeling [4]. Although these methods are useful in maximizing parallelism, they do not aggregate independent computations into regions with fixed sizes and regular shapes.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Tiling Nested Loops into Maximal Rectangular Blocks

Yeong-Sheng

Wang

1996

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

Section: B Motivationmentioning

confidence: 99%

Section: A Terminologymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Tiling Nested Loops into Maximal Rectangular Blocks

Yeong-Sheng

Wang

1996

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

“…In order to minimize the amount of communication overhead, several researchers have tried to partition iterations of nested loops into independent blocks so that there is no data dependence relation among the partitioned blocks [D'Hollander 1989;Peir and Cytron 1989;Shang and Fortes 1988]. However, in practice, many programs cannot be partitioned into independent blocks by applying such partitioning methods; thus these programs will be executed sequentially.…”

Section: Introductionmentioning

confidence: 99%

Partitioning and mapping of nested loops for linear array multicomputers

Sheu

Chen

1995

J Supercomput

View full text Add to dashboard Cite

Abstract. In distributed-memory multicomputers, minimizing interprocessor communication is the key to the efficient execution of parallel programs. In order to reduce the amount of communication overhead, parallel programs on multicomputers must be carefully scheduled by pamllelizing compilers. This paper proposes some compilation techniques for partitioning and mapping nested loops with constant data dependences onto linear array multicomputers. First, a systematic partition strategy is proposed to project an n-dimensional computational structure, representing an n-nested loop, onto a line to form a one-dimensional projected structure with low communication overhead. Then, a mapping algorithm is proposed for mapping the partitioned loops onto linear arrays in a way that balances the workload and minimizes the communication cost among processors. Finally, parallel execution codes can be automatically generated for such linear array multieomputers.

show abstract

“…Pure systolic architectures happen to be impractical since they exhibit too much parallelism to meet the bandwidth and resource constraints. Strategies were proposed to derive processor arrays in the presence of resource or I/O constraints [20,15,12,5,4,6], leading to a deep understanding of the partitioning transformation [21,9,6,8].…”

Section: Introductionmentioning

confidence: 99%