Sequential decomposition of operations and compilers optimization - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Rapport (Rapport De Recherche) Année : 2009

Sequential decomposition of operations and compilers optimization

Résumé

Code optimization is an important area of research that has remarkable contributions in addressing the challenges of information technology. It has introduced a new trend in hardware as well as in software. Efforts that have been made in this context led to introduce a new foundation, both for compilers and processors. In this report we study different techniques used for sequential decomposition of mappings without using extra variables. We focus on finding and improving these techniques of computations. Especially, we are interested in developing methods and efficient heuristic algorithms to find the decompositions and implementing these methods in particular cases. We want to implement these methods in a compiler with an aim of optimizing code in machine language. It is always possible to calculate an operation related to K registers by a sequence of assignments using only these K registers. We verified the results and introduced new methods. We described In Situ computation of linear mapping by a sequence of linear assignments over the set of integers and investigated bound for the algorithm. We introduced a method for the case of boolean bijective mappings via algebraic operations over polynomials in GF(2). We implemented these methods using Maple
Fichier principal
Vignette du fichier
RR-7076.pdf (634.14 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

inria-00428722 , version 1 (29-10-2009)

Identifiants

  • HAL Id : inria-00428722 , version 1

Citer

Mumtaz Ahmad, Serge Burckel, Adam Cichon. Sequential decomposition of operations and compilers optimization. [Research Report] RR-7076, INRIA. 2009, pp.91. ⟨inria-00428722⟩
435 Consultations
400 Téléchargements

Partager

Gmail Facebook X LinkedIn More