On the density and discrepancy of a 2D point set with applications to thermal analysis of VLSI chips

Abstract

In this era of giga-scale integration, thermal analysis has become one of the hot topics in VLSI chip design. Active thermal sources may be abstracted as a set of weighted points on a 2D chip-floor. The conventional notion of discrepancy that deals with the congestion properties of a set of scattered points may not be able to capture properly all real-life instances in this context. In this paper, we have introduced a new concept, called the density of a region to study some of the properties of the distribution of these weighted points. We prove several counter-intuitive results concerning the properties of the regions that have maximum or minimum density. We then outline algorithms for recognizing these regions. We also compare the attributes of density with the existing concept of discrepancy.