System design has entered a period of rapid change that is shaking up and reconfiguring established companies and entire industry sectors. At the core of this change is the trend towards the electrification and digitization of … almost everything. Consider things like cars, shopping, and telephones that were essentially devoid of complex electronics a few decades ago and are now intensely dependent on digitization and computers. Adding fuel to this fire is the explosive adoption of artificial intelligence, machine learning, and digital twins by many industries. The result is that designing a system today is significantly more complicated and pulls together multiple branches of physics expertise.
Automotive and other industries are going through seismic upheavals as they rush to adapt their design abilities to include semiconductors, battery technology, and autonomous driving. Jet engine manufacturers and big-iron industrial suppliers like Siemens all offer digital twin technology for system lifetime management. And in the electronic design sector, the slowdown of Moore’s Law along with the rise of 2.5D and 3D packaging is driving the consolidation of traditionally separate disciplines like chip design, package design, and printed circuit board (PCB) design. Or as Forbes put it in a recent article by Tim Bajarin, “Welcome to the newest phase where packaging starts to matter more than process technology.”
The siloed isolation of chip design from package design and PCB design means that each of these markets has developed insular data structures that are ill-suited to deal with the breadth of multiphysics analyses needed for 3D-IC design. We see this same issue in the broader engineering ecosystem where many different physical disciplines, including computational fluid dynamics, mechanical stress, electromagnetic radiation, and materials science, all need to work together. I believe a critical foundation for success in this new multiphysics reality is the development of open, extensible and cloud-optimized platforms that enable many different tools and data sets to work together and build towards a realistic and accurate simulation of physical reality.
The platforms we are building need to be open because no one company has all the technology to cover the broad range of today’s engineering challenges. Here at Ansys we sometimes say that our engineering simulation technology extends “from chips to satellites,” which captures the scope of the challenges we are aiming to solve. By open I am not implying that the platforms must be built with open-source software – that is a different topic. Open here refers to the ability to freely exchange data with other tools and other platforms. The opposite of the closed garden approach to platforms. A successful example of such an open platform is the Virtuoso™ full-custom design platform from Cadence Design Systems. It has enabled a broad ecosystem of specialized third-party tool providers to come together in what has been a win-win-win for Cadence, the tool providers, and the design community.
The scope of multiphysics simulations exceeds not only the product offerings of a single company but also varies tremendously by customer application. An essential component of any successful platform is the ability for users to customize and extend its capabilities. High-performance API access functions and popular programming languages like Python let customers bridge the gap between a standard product and an optimized solution for them. For example, a key factor in the success of Virtuoso is its powerful SKILL programming language, while Synopsys’ PrimeTime offers a widely used extension language for timing analysis and flow integration.
The third key ingredient for successful platforms of the future is a fundamental embrace of the distributed cloud computing paradigm. A consequence of the growing complexity of today’s engineering challenges is that they need enormous amounts of compute power for certain critical tasks. This is an ideal scenario for working in the cloud and the platform must support and enable this, not as an add-on afterthought, but as its fundamental operating mode.
In summary, I firmly believe that system design in the coming years will only succeed in fulfilling its promise if it is supported by a new generation of open, extensible, and cloud-enabled platforms. Here at Ansys we have a head start with over a decade of investment in platform technologies like SeaScape™ and Ansys Engineering Desktop (AEDT™) that are purpose-built to enable multiphysics simulations. We are excited about taking these platforms to the next level in enabling the design community to automate and optimize complex workflows and solutions, not only between products and databases within Ansys but also those outside of Ansys, with initiatives such as the recent availability of Python-based pyAEDT on GitHub.
About the Author
John Lee is vice president and general manager of the Ansys Electronics, Semiconductor, and Optics Business Unit. Lee co-founded and served as CEO of Gear Design Solutions (now Ansys), developer of the first purpose-built big data platform for integrated circuit design. He cofounded two other startups (Mojave Design and Performance Signal Integrity), which successfully exited into companies now part of Synopsys. He holds undergraduate and graduate degrees from Carnegie Mellon University.