• DocumentCode
    2857722
  • Title

    Array Data Dependence Testing with the Chains of Recurrences Algebra

  • Author

    Van Engelen, Robert A. ; Birch, Johnnie ; Gallivan, Kyle A.

  • Author_Institution
    Sch. of Comput. Sci. & Inf. Technol., Florida State Univ.
  • fYear
    2004
  • fDate
    12-14 Jan. 2004
  • Firstpage
    70
  • Lastpage
    81
  • Abstract
    This paper presents a new approach to dependence testing in the presence of nonlinear and non-closed array index expressions and pointer references. The chains of recurrences formalism and algebra is used to analyze the recurrence relations of induction variables, and for constructing recurrence forms of array index expressions and pointer references. We use these recurrence forms to determine if the array and pointer references are free of dependences in a loop nest. Our recurrence formulation enhances the accuracy of standard dependence algorithms such as the extreme value test and range test. Because the recurrence forms are easily converted to closed forms (when they exist), induction variable substitution and array recovery can be delayed until after the loop is analyzed
  • Keywords
    optimising compilers; system recovery; array data dependence testing; induction variables; nonclosed array index expressions; nonlinear array index expressions; pointer references; recurrences algebra; recurrences formalism; Algebra; Arithmetic; Benchmark testing; Computer science; Delay; Information technology; Parallel processing; Performance analysis; Polynomials; US Department of Energy;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Innovative Architecture for Future Generation High-Performance Processors and Systems, 2004. Proceedings
  • Conference_Location
    Maui, HI
  • ISSN
    1537-3223
  • Print_ISBN
    0-7695-2205-X
  • Type

    conf

  • DOI
    10.1109/IWIA.2004.10018
  • Filename
    1410682