Practical Chip Design

Intel's Ken Tallo has taken up the chairman's baton at the Open SystemC Initiative, the organization that drives adoption, infrastructure, and support for SystemC. Speaking with Tallo yesterday, we had a glimpse into one insider's view of the future of the organization, the language, and to some degree, the whole technology of system-level behavioral modeling.

At one level, Tallo's aspirations are what would be expected of any incoming chairman: increase membership, improve the quality of the organization's functions, and strengthen relationships with allied organizations. But once we moved from generalities to specifics, Tallo's view of the system modeling world became quite fascinating.

Let's start with broadening the membership. In Tallo's view, this is not a matter of rounding up the remaining second- and third-tier EDA companies and IDMs and herding them into the membership at reduced rates. Nor is it even a given that the organization should grow organically. Rather, increasing membership is in Tallo's opinion a consequence of a naturally broadening base for system modeling as a discipline.

"SystemC is becoming valuable for all sorts of virtual platforms, not just for digital ICs," Tallo said. The most obvious extension is into mixed-signal functionality, a program that OSCI already has underway with the Analog/Mixed-Signal proposal now out for comment. So one issue in broadening membership is to draw in the analog-specialty companies that have traditionally not been involved in SystemC activities. "It is becoming as important to analog houses as to digital designers to model and explore the interfaces between analog and digital functions," Tallo said. "So system-level, mixed-signal modeling is becoming an important capability even to teams that have mastered analog design."

But this is just the beginning of the future. Beyond analog modeling lie RF, package, and interconnect modeling, also increasingly important to getting system-level functionality right. None of these parts of the system can be left as implementation details in high-performance design. Beyond this lie other views of the system than increasingly must be treated at the system level: power, heat, and even mechanical behavior. "It's just a matter of time until the discipline broadens to encompass other applications," Tallo said.

Along another axis altogether, there is the ever-present question of synthesis, still the eventual aim of many SystemC users. "It is always our desire to bridge that void between behavioral modeling and RTL or gate-level design," Tallo said. "In fact, we have seen some very interesting progress in the last year, but this is still an area where the technology is maturing."

Broadening the membership beyond chip designers, and perhaps even beyond electronic design. Creating a language that can model an entire system, not just the electronic behavior of the chips. Moving toward synthesis directly from the system-level view. Those are all aspirations that will challenge the organizational skills of OSCI and its leadership as much as they will test the technical abilities of the engineers and researchers in the working groups.