Toward CAD-IP Reuse: The MARCO GSRC Bookshelf of Fundamental CAD Algorithms

Free reuse of code and other types of intellectual property in CAD can be in the mutual interests of commercial EDA vendors, captive CAD organizations, and academic researchers as they together address the "Design Technology Productivity Gap". We discuss the productivity and cultural needs that motivate a community-wide emphasis on CAD-IP reuse, then de ne what constitutes CAD-IP. We describe a speci c framework for CAD-IP reuse, an instance of which is now operational (under the auspices of the MARCO Gigascale Silicon Research Center) and accessible to the general public over the Internet. 1 The Design Technology Productivity Gap and the Need for Reuse Future requirements for design technology are always uncertain due to accelerating changes in process technology, system implementation platforms, and electronics markets. The \time-to-market" for new design technology { from understanding the design problem to integration of a new solution in mainstream ows { can span several process generations and the birth and death of entire electronics markets. Thus, it is di cult for designers to obtain the right design technology at the right time. Overall, there is a great need for the entire design technology community (commercial vendors, captive CAD organizations, and academic researchers) to focus on improving the delivery of design technology. Put another way, there is a need to at least partially address the well-known \Design Productivity Gap" by addressing a less well-known \Design Technology Productivity Gap". The design technology landscape today reveals many indicators of a design technology productivity gap. For example, at a high and somewhat abstract level, phrases such as \design productivity gap" and the very existence of the MARCO Design and Test Focus Research Center re ect a perceived need to improve the e ectiveness and impact of CAD algorithms research. At a more technical level, fragmentation of research across too many subcommunities creates a need to regain awareness of and focus on a well-de ned \leading edge". Here, we note that the 1999 Design Automation Conference instituted a new topic area, T0.1: Fundamental CAD Algorithms. And at the level of individual problem formulations (speci cally, hypergraph bipartitioning), Caldwell et al. [10] noted cases of silent, undocumented implementation decisions that can change quality of result by 400+%, and the existence of two papers published since April 1998 that report 1000+% di erences in solution costs returned by implementations of the same well-known algorithm. Why Reuse? The standard criteria for technology delivery are time-to-market and quality-of-result achieved within given resource constraints. We observe that all of these criteria strongly motivate a culture of reuse throughout the design technology community: Supported in part by the MARCO/DARPA Gigascale Silicon Research Center and by a grant from Cadence Design Systems, Inc. Author email contact information: [email protected], [email protected], [email protected]