Heat transfer research for cooling electronic equipment has to be planned and conducted on firm understanding of working principle of the computing system and inherent constraints arising thereof. The present paper outlines the computing system’s requirements, and explains some research topics which will be important for the development of future generations of computers. Major points of presentation are as follows. (1) The progress in circuit miniaturization that has been achieved in the past decade will have to be slowed down in the near future due in large part to the rising cost of circuit fabrication and the difficulty of chip cooling under constraints pertaining to electronic systems. The latter factor points to the possibility that heat flux on the chip will not be a dominant factor in further improvement of the computer performance as it has been by now. Heat transfer research focusing only on cooling chips will, therefore, deliver less impacts on the future course of computer technology. (2) The demand to minimize communication delay time within the system, hence to make system-level wiring shorter, sets the preference for system morphology, and the geometrical configuration of the system defines optimum cooling scheme. The result of the case study for an immersion-cooled card-stack processor revealed that the optimum coolant channel spacing is in a sub-millimeter range. Heat transfer to fluorinert flowing in a channel of such small dimensions is an important research subject. In particular, two-phase flow heat transfer in sub-millimeter channels needs to be a focus of future heat transfer research. (3) Downsizing trend in the computer industry is urging equipment designers to exercise thermal management also in system-level perspectives. A thermal/system-volume trade-offs is illustrated using a simplified model. The study shows that a 10 % increase in the cabinet volume to make room for throughflow of cooling air appreciably improves the thermal environment in the cabinet. (4) Heat transfer from a localized heat source in an air-cooled downsized computer presents a challenging research topic due to its conjugate nature. The experiment is conducted mounting a square module on a flat copper plate in a parallel-plate channel. Complex physical process in this geometrically simple system is outlined, and some examples of the heat transfer data are shown.
[1]
Wataru Nakayama,et al.
The Effect of Module / Board Interface Thermal Resistance on Cooling of an Air-Cooled Module
,
1993
.
[2]
R. Phillips,et al.
Microchannel heat sinks
,
1988
.
[3]
Robert W. Keyes,et al.
The physics of VLSI systems
,
1987,
Microelectronics systems design series.
[4]
B. Prince,et al.
ICs going on a 3-V diet
,
1992,
IEEE Spectrum.
[5]
Frank P. Incropera,et al.
Conjugate heat transfer from small isothermal heat sources embedded in a large substrate
,
1985
.
[6]
D. K. Ferry.
Interconnection lengths and VLSI
,
1985,
IEEE Circuits and Devices Magazine.
[7]
William E. Donath,et al.
Placement and average interconnection lengths of computer logic
,
1979
.
[8]
W. Nakayama.
A data shifter-the vehicle for study of thermal-electrical interactions in integrated circuits
,
1992,
[1992 Proceedings] Intersociety Conference on Thermal Phenomena in Electronic Systems.
[9]
Wataru Nakayama.
Computers and Computing in Heat Transfer Science and Engineering
,
1992
.
[10]
Raul Mendez.
High performance computing: research and practice in Japan
,
1992
.
[11]
J. Whitelaw,et al.
Convective heat and mass transfer
,
1966
.
[12]
H. B. Bakoglu,et al.
Circuits, interconnections, and packaging for VLSI
,
1990
.