Why matrix reduction is better than objective function based optimization in compact thermal model creation

Abstract

It is shown how the objective function optimization approach fails in the construction of boundary condition independent compact thermal models of complex Systems On Silicon in which more than two powers are allowed to arbitrarily change. For such applications a novel matrix reduction approach is shown to lead to the efficient construction of accurate boundary condition independent compact thermal models whichever is the number of independent heat sources. Such matrix reduction approach is implemented in the FANTASTIC BCI software tool which is allow to analyze state-of-the art industrial applications.