Skip to Main content Skip to Navigation

Etude d'un coeur de processeur pour l'arithmétique exacte

Abstract : The floating point arithmetics used within machines for scientific calculations introduce errors within the results of such processings. Indeed computer processings consider a limited number of significant figures for operands, thereby corresponding to only an approximation of the exact value. As programs get executed, there is a gradual loss in the precision of the numbers handled. These accumulations of errors may lead to incorrect results without the user being aware of it. Among the solutions developed in order to take into account these calculation errors, only exact arithmetics lead to a result on which one may thrust. It is thus necessary to emulate each elementary operation of exact arithmetics using software routines according to the statements available on the processor. Such an emulation gives rise to a loss in the arithmetic precision and thus in that of the software used for the exact calculation. We propose to develop a processor that would execute all the elementary operations of exact arithmetics in hardware. The architecture of the circuit will be optimised in order to handle the characteristics of such arithmetics and to allow the processing of large numbers. Moreover to increase the performances of software, some common functions of exact calculations will be implemented in hardware.
Complete list of metadatas

Cited literature [73 references]  Display  Hide  Download
Contributor : Lucie Torella <>
Submitted on : Friday, June 13, 2003 - 3:43:48 PM
Last modification on : Friday, December 11, 2020 - 8:28:03 AM
Long-term archiving on: : Friday, April 2, 2010 - 8:09:58 PM


  • HAL Id : tel-00002997, version 1




V. Coissard. Etude d'un coeur de processeur pour l'arithmétique exacte. Micro et nanotechnologies/Microélectronique. Institut National Polytechnique de Grenoble - INPG, 2000. Français. ⟨tel-00002997⟩



Record views


Files downloads