Acceleration of engineering and scientific applications using trivial operand processing

Abstract

A vast array of scientific and engineering problems require the solution of linear systems of equations of the form A* x = y , where A is the coefficient matrix of the system and y is a vector of unknowns and jc is a vector of scalar known values. In practice the matrix A is large and sparse for real-world problems. A range of iterative methods is used depending on the nature of the problem to be solved and multiple solvers are normally included in commercial and public domain applications based on these methods. Most implementations of iterative methods [3] and the linear algebraic operations on which they are based have to date been software implementations commonly implemented in the form of FORTRAN or C/C++ libraries. All mathematical libraries in common use make usee of data-structures to store sparse matrices efficiently, and such methods can thus be regarded as a form of compression.