SOC designers today not only have to design at denser geometries and with greater complexities, but also have to manage critical housekeeping chores such as well-tracking dependencies, controlling costs, and ensuring design jobs are performed in the proper order. Manual techniques no longer suffice in managing such complexity.
Proper flow management of the SOC design process is key to competitiveness in the marketplace, reduces costs, and brings products to market faster. To manage such designs, engineering departments must search for ways to maximize resources by seeking new and improved solutions.
An optimized automatic system for flow management at the operational level-at the level of the dependencies between files and tools-can address these issues. Flow management streamlines the entire design process from concept to silicon, while maximizing resource (CPUs and software licenses) and providing a single, consistent structure for managing the design flow. In turn, the design team is able to integrate disparate tools from multiple vendors, reduce the number of CPU cycles required by avoiding tool executions, and shorten overall time to market. As a result, engineers gain more control over the entire development process, including planning, synthesis, verification, and testing.
As SOC designs are growing more complex, the designs become more difficult to track. Complex designs require more of the designer's time and more support. Jobs take longer to run and may have to be re-run, tying up valuable CPU cycles and software licenses. The final design may still contain major flaws.
Efficiencies in all these areas are currently being overlooked. By using an optimized system for flow management, the design team can uncover a motherlode of productivity-design teams can run more jobs in less time. Flow management also gives engineers more control over the entire job.
Because flow management allows engineering departments to maximize the utilization of existing CPUs and software licenses, some EDA vendors may perceive flow management as a threat to their ability to sell more licenses. On the contrary, my experience is that flow management improves resource utilization while it promotes the use of more resources, such as CPUs and licenses.
A routine sub-project involving library characterization can have anywhere from 10,000 to 100,000 files, or more, and tens of thousands of jobs. Who can say whether all files are up to date and all measurements performed successfully? The lack of a complete representation of the relationships between the files leaves open the possibility that some changes may not propagate to all dependent files and errors may remain undetected. Using an automatic system designed to track the flow and the interdependencies between the files and the tools, designers can correctly propagate the changes to the parts of the design that depend on the changed file.
Flow management also improves the interoperability of the tools. Of course, there are dozens of excellent specialized tools designed for each of the various stages of the design process. Flow management provides a single, cohesive structure for ensuring that all of these tools work together at maximum efficiency. In this way, flow management increases the probability that all jobs run with valid, current data that reflect the latest design changes. In addition, to increased utilization of resources, flow management can dramatically improve designs. Greater efficiency also results in more competitive designs and faster time to market.
Flow management, which focuses on the operational relationship between files and tools, returns control of the process to the designer while easing the burden of design integrity. For the SOC designer, remaining competitive means continually striving for ways to reduce costs and shorten time to market. Flow management is one technique that works.
Andrea Casotto is president and founder of Runtime Design Automation, founded in 1995. Prior to Runtime, Casotto was with Siemens Corporate Research in Munich, Germany. He received his Ph.D. in electrical engineering from U.C. Berkeley.