Parallel loops — A test suite for parallelizing compilers: Description and example results

Dongarra, Jack; Furtney, Mark; Reinhardt, Steven K.; Russell, J.R.

doi:10.1016/s0167-8191(05)80036-5

Cited by 25 publications

(8 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We tested the I test and the non-continuous I test and performed experiments on the codes abstracted from the following four numerical packages: Vector Loop, Livermore, MDG (Perfect Benchmarks), and MG3D (Perfect Benchmarks) [8,32,33]. One-hundred and forty pairs of onedimensional array references were observed to have subscripts with non-singular increments.…”

Section: Resultsmentioning

confidence: 99%

“…Where and are integers for and and are lower bound, upper bound, and the increment of a general loop, respectively, and P is the number of loop iterations in the general loop and P = The GCD test, the Banerjee test, and Fourier-Motzkin elimination are three basic dependence analysis techniques but are too naive or expensive in practice [3,[8][9][10][11]. There have been various advanced techniques to extend the above methods for overcoming the disadvantages of them [12][13][14][15][16][17][18].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

The Accuracy of the Non-continuous I Test for One-Dimensional Arrays with References Created by Induction Variables

Zhang¹

2014

Journal of Information Processing Systems

View full text Add to dashboard Cite

Abstract-One-dimensional arrays with subscripts formed by induction variables in real programs appear quite frequently. For most famous data dependence testing methods, checking if integer-valued solutions exist for one-dimensional arrays with references created by induction variable is very difficult. The I test, which is a refined combination of the GCD and Banerjee tests, is an efficient and precise data dependence testing technique to compute if integer-valued solutions exist for one-dimensional arrays with constant bounds and single increments. In this paper, the non-continuous I test, which is an extension of the I test, is proposed to figure out whether there are integer-valued solutions for one-dimensional arrays with constant bounds and non-sing ularincrements or not. Experiments with the benchmarks that have been cited from Livermore and Vector Loop, reveal that there are definitive results for 67 pairs of onedimensional arrays that were tested.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

The Accuracy of the Non-continuous I Test for One-Dimensional Arrays with References Created by Induction Variables

Zhang¹

2014

Journal of Information Processing Systems

View full text Add to dashboard Cite

show abstract

“…This study applied the Quadratic test and performed experiments in the parallel tool on a personal computer equipped with an Intel Pentium III CPU using the codes selected from five numerical packages -EIS-PACK, LINPACK, Parallel loops, Livermore loops and Vector loops [3,15,27]. The codes have more than 37,000 lines of statements, 610 subroutines and 3663 loops.…”

Section: Resultsmentioning

confidence: 99%

An exact data dependence testing method for quadratic expressions

Chu

2007

Information Sciences

View full text Add to dashboard Cite

“…We tested the multi-dimensional generalized interval test and performed experiments on Personal Computer Intel Pentium IV through the benchmark codes cited from [11,18,28]. 515 pairs of array references were observed to have linear subscripts.…”

Section: Resultsmentioning

confidence: 99%

A Polynomial-Time Dependence Test for Determining Integer-Valued Solutions in Multi-Dimensional Arrays Under Variable Bounds

Chang

Chu²,

Wu³

2005

J Supercomput

View full text Add to dashboard Cite

Multi-dimensional arrays with linear subscripts occur quite frequently in real programs. For multidimensional linear arrays under variable bounds as well as any given direction vectors, the generalized Lambda test is an efficient and precise data dependence method to check whether there exist real-valued solutions. In this paper, we propose a multi-dimensional generalized interval test-a polynomial-time dependence test that can be applied towards testing whether there are integer-valued solutions for multi-dimensional linear arrays under variable limits and any given direction vectors. Experimental results with benchmark showing the effects of the multi-dimensional generalized interval test over the generalized Lambda test are also presented.

show abstract

Parallel loops — A test suite for parallelizing compilers: Description and example results

Cited by 25 publications

References 1 publication

The Accuracy of the Non-continuous I Test for One-Dimensional Arrays with References Created by Induction Variables

The Accuracy of the Non-continuous I Test for One-Dimensional Arrays with References Created by Induction Variables

An exact data dependence testing method for quadratic expressions

A Polynomial-Time Dependence Test for Determining Integer-Valued Solutions in Multi-Dimensional Arrays Under Variable Bounds

Contact Info

Product

Resources

About