While engineering design tools continue to improve, the products and systems these tools are called upon to design continue increasing in complexity, with higher operating frequencies, greater density, and more advanced parts. The key to tackling these challenges will be the incorporation of AI into the design flow, according to Ben Gu, Vice President of Research and Development at EDA vendor Cadence Design Systems, of Cadence.
Gu echoed his thoughts on intelligent system design at a session titled “The Intelligence to Design Intelligent Machines,“ during the recent DesignCon Conference and Exhibition in Santa Clara, CA, Gu said that AI-inspired design tools would give engineers the added brain power to speed calculations and navigate through challenging design issues. He believes AI would help increase team productivity, improve the quality of results, and improve engineering design throughput.
Cadence, along with other EDA and engineering design tool vendors, are in the midst of a massive upgrading of their design software to incorporate machine learning algorithms. These algorithms are enabling the design tools to synthesize massive amounts of data and derived optimized potential outcomes far faster than with previous design.
“In the past, engineers would have to input system parameters, run simulations, then analyze and change parameters, in a number of design iterations over and over until optimal results are achieved,” Gu said. “Machine-driven AI can achieve faster simulation and optimize results faster.”
AI Aids Debugging to Reduce Costly Errors
But speed is just part of the equation. Gu said that AI-driven design can significantly increase productivity in debugging designs, as bugs inevitably surface in design iterations.
“Logic verification imposes some of the highest costs,” Gu noted. “Late bugs can cost exponentially more, particularly as we put more transistors on a chip and the number of gates reaching staggering levels.”
Gu added, “IP is constantly changing, and weekly SoC testing is resulting in many failures. Finding the root cause of failures is time consuming.”
Gu cited several examples of products where AI helped to improve design productivity. “One was a mobile CPU built on a 5-nm process, where a improved, converged design flow was achieved within 10 days compared to typically many weeks or months. In another case, AI was able to reduce power needs an additional 5% for a high-performance GPU, in addition to improving design productivity.”
AI is not only helping to speed tasks such as verification, simulation, and layout, but also system planning and completion of final tasks, according to Gu. “AI can optimize the concept, implementation, and signoff, achieving intelligent system optimization.” Gu added that AI will be able to use machine learning to build better models for future designs, which will further improve design productivity.
Spencer Chin is a Senior Editor for Design News covering the electronics beat. He has many years of experience covering developments in components, semiconductors, subsystems, power, and other facets of electronics from both a business/supply-chain and technology perspective. He can be reached at [email protected]