Design automation in MEDEA: present and future

0272-1732/99/$10.00  1999 IEEE The semiconductor industry has been growing at an unprecedented level since its start in the early 1960s. Capitalizing on the outstanding properties of silicon and its stable oxide permitted the introduction of the CMOS industry, the leading semiconductor industry process. However, the average 15% to 16% annual growth in semiconductor sales has also presented tremendous problems of huge investments in manufacturing. Mandatory now is a rapid return of investment through advanced products (in the latest available processes) that have high added value at the system level. Sometimes the system is the product itself, which raises thoughts of new ways to design these complex systems on a single chip that mixes several functionalities. This article explores a future European evolution of design automation, discussing the present status of Europe’s lack of local industrial developments as well as its tremendous knowledge reservoir. Earlier MEDEA activities paved the road to new design solutions and gave European companies the chance to influence US developments, as is evidenced by frequent technology partnerships with US software vendors. Recently, we have seen a significant increase in European start-ups in advanced design automation domains (hardware-software codesign, intellectual property reuse, deep-submicron effects). Recommendations Research has generally analyzed the evolution to system-on-chip capabilities through silicon process evolution capabilities, an enabler at the same level as design automation. The end product is the silicon chip itself that sells on the market. The early accessibility of products in the latest silicon is key for the maturity of these processes. Although Europe lags behind the US here, process development is often synchronous. One of Europe’s strategic objectives should be to speed its contribution of engineering the design solutions in various silicon application platforms of choice. This will enable companies to develop early, reusable system intellectual properties for the next product generation. These system intellectual properties already include basic digital functions with addedvalue functionalities such as analog (a longlasting, well-recognized European expertise), RF, embedded memories (SRAM, DRAM, nonvolatile), and micromechanical functions. Some basic recommendations must be reassessed. One, time to market with systemon-chip products, is a killer parameter. In the future, we will see one silicon manufacturer instead of a shared market covered by several older dies. Figure 1 shows the impact of being late in time to market versus having access to the whole product market and being competJoseph Borel