Heat-transfer engineering in systems integration: outlook for closer coupling of thermal and electrical designs of computers

This paper begins with a review of the author's personal experience in the research field of computer cooling. It highlights the need to develop foresight on the possible course of hardware development in order to provide the package designer with appropriate heat-transfer data in a timely manner. A question is then raised about the immediate future of the (indirect) water-cooling technology. Water-cooling has so far proven effective in cooling high-end computers which use ECL devices in two-dimensional packaging. The drive toward higher raw speeds of ECL devices, however, is going to lose steam-emerging instead is the endeavor to upgrade system performance by massively-parallel computing which requires wiring-intensive hardware. Three-dimensional packaging will meet the demand for short global wiring in systems, but will become a commercial reality only after the establishment of methodologies for its design and assembling. One of the key issues in the design of 3-D computers is the optimum allocation of physical space for electrical wiring and heat-transfer paths. Intimate coupling of wiring and heat transfer designs pose challenges to heat-transfer researchers that have not surfaced in other industrial applications. Items of primary importance include: the methodology to predict how and temperature distributions in a field having a wide spectrum of length scales, the local heat-transfer coefficients in the maze of microscale coolant channels, the possibly large effect of extraneous factors such as irregular geometric features of coolant channels and conjugate mode of heat transfer, and temperature control during assembling of 3-D structures.