July - August 1999
eMail the Editor
Gigascale Integration Will Transcend Current Interconnection LimitsBy Dr. Kevin P. Martin
Microelectronics Research Center, Georgia Institute of Technology, Atlanta
Left unchanged, these GSI interconnect limits would otherwise slow or halt the historical rate of progress of gigascale integration.
This 39-month, $19.5 million IFC program is jointly funded by the Defense Advanced Research Project Agency (DARPA) and the Microelec-tonics Advanced Research Corp. (MARCO), which is funded by the Semiconductor Industry Association, San Jose.
Georgia Tech is the prime contractor and lead institution in the multi-university research consortium, which also includes Cornell, MIT, Rensselaer, Stanford and SUNY-Albany. Prof. James D. Meindl, director of the Georgia Tech Microelectronics Research Center is IFC director.
Virtually all contemporary architectures for microprocessor and logic chips are derived from currently obsolescent, original implementations whose performance and cost were dominated by transistor properties, as opposed to interconnect properties.
Consequently, a key forward looking feature of the IFC program will be a research effort dealing with interconnect-centric micro-architectures. Core tasks will deal with materials and processing, process modeling and simulation and reliability and characterization.
Early 21st century opportunities for gigascale and terascale integration will be governed by a hierarchy of physical limits with five levels. These are:
The most severe limits in this hierarchy will not be imposed by transistors performing computing functions, but rather by interconnection networks broadly conceived as performing signal communication, clock distribution and power distribution functions.
In this context, fundamental, material, device, circuit and system limits, without exception, dictate the wiring imperative to "keep interconnects short" to reduce response time-as well as power dissipation, signal contamination and failure rates-to the smallest possible values.
The seven cross-cutting research tasks that the program will pursue are:
New frontier projects will begin immediately during the first year of IFC operation at each of the six initial universities.
These projects will be augmented in following years, both by adding new universities to the IFC and by expanding the number and scope of new frontier projects at the six initial universities.
To facilitate timeliness and efficiency in transferring results, the IFC has established a WWW site (http://ifc.gatech.edu) whose contents will be refreshed at least quarterly.
The management system is designed to maximize center-wide synergy via a novel strategy. Vertical or interdisciplinary synergy within a single institution will be stimulated by the Leadership Council member representing that institution; horizontal or intra-disciplinary synergy within a single research task will be stimulated by the task leaders).
This tightly interwoven matrix management system offers the proposed IFC rare opportunities for synergy not available in more common academic and industrial research settings.
Readers may contact Dr. Martin at firstname.lastname@example.org or phone 404.894.4035.
Chip Scale Review o 7291 Coronado Drive, Suite 8 o San Jose, CA 95129 o Email: email@example.com
|© 1998 ChipScale REVIEW|