To better understand the evolution of physical design, one might compare it with the evolution of mankind. The similarities can help us understand where the next breakthrough might come from.
The earliest automated design implementations were performed in a sequential flow. This Homo Habilis¹ approach used synthesis tools to speed up design implementation. Although cumbersome, the implementation flow included automation. Tools were primitive but sufficient.
The next stage in tool evolution, Homo Erectus², allowed engineers to stand up. Tools advanced to support diversified requirements. The flow was still sequential, and each tool was dedicated to its purpose. Over-the-wall placement, routing, and sign-off were separate stages with little to no interaction, similar to the "simple" tools used by "standing man." As design implementation evolved, this "over-the-wall" approach no longer sufficed, and the flow required more advanced tooling for design implementation.
Homo Neanderthalensis³ tools were innovative, multipurpose, and advanced, permitting optimization of more than one goal. Placement had evolved to concurrent analysis and optimization: physical synthesis.
Moving to the next step in evolution, Homo Sapiens4; integrated the tools (computers, cell phones, PDAs, and others) being used. The equivalent technological advance—interconnect synthesis—melds optimization into routing while progressively optimizing routing, placement, and netlist to achieve electrical and geometrical convergence.
What is the next step in the evolution? Homo Futurus? Adding a new tool, such as a new analysis or optimization engine or a new PDA, is not an evolutionary step. Instead, we are searching for the new "Internet" of physical design. What will this methodological change entail? Clearly, it will include an integrated approach, combining old with new. Already, manufacturing constraints are impacting the implementation flow.
Design for manufacturing (DFM) must be addressed as an implementation step that considers electrical implications. It can't be treated as a new standalone tool or flow. The old school of physical design still requires electrical closure in addition to DFM optimization.
Conflicting rules demand a new approach, resulting in the next step in evolution—DFM interconnect synthesis—a geometrical and electrical solution that melds DFM optimization into the electrical and geometrical closure. DFM optimization must be addressed during design convergence because geometrical closure, electrical closure, and DFM closure are tightly interdependent.
1 Homo Habilis: called "handy man" because of evidence of tools found with its remains.
2 Homo Erectus: "standing man." There is evidence that Erectus probably used fire. Their stone tools were more sophisticated than Habilis tools.
3 Homo Neanderthalensis: Many of the tools and weapons found were more advanced than those of Homo Erectus.
4 Homo Sapiens: Modern man uses integrated tools and advanced technologies to combine multiple tasks.
Jacob Greidinger is CEO and vice president of R&D at Silicon Design Systems. He is a 25-year EDA veteran with stints at many vendors. Also, he holds three U.S. patents and has filed four additional patents.