Funding, design scale, and access to the latest industry technologies to enable cutting-edge research are a challenge for many universities. University research is still important today, but less critical for core EDA. Many of the recent EDA advances have come directly from the internal development efforts of EDA companies rather than from academia. However, we do see academic research as increasingly important for future technology advancements in areas such as next-generation lithography, new transistor channel material, and new device structures.
Why are there fewer EDA advancements coming out of universities today? Major sources of academic research funding have diverted their attention away from EDA and toward more global challenges such as energy, sustainability, and health, resulting in fewer professors and students in the field of EDA. Additionally, many top engineering graduates are gravitating to Google and other high-visibility companies rather than EDA. With less funding and focus on EDA, it is difficult for academia to develop unique solutions on their own. Industry collaboration with academia is one way to fill the gap, and we expect to see this trend increase over time.
EDA| Cadence, Mentor and Synopsys, the three largest EDA companies, were all founded nearly 25 or more years ago. What do you think has been EDA’s biggest innovation in that time?
Logic synthesis, the technology upon which Synopsys was founded, was revolutionary. The transition from schematics to language-based design enabled design engineers to achieve a tremendous productivity gain and changed IC design permanently for the better. While this was perhaps EDA’s biggest innovation, the history of EDA has been characterized by many other significant
- Physical synthesis enabled synthesis to maintain accuracy as wires became dominant.
- Static timing analysis was fundamental to designing large blocks without requiring exhaustive timing patterns.
- Formal verification enabled incremental optimization without exhaustive re-verification.
- The advent of IP blocks from reliable vendors provided the industry’s biggest productivity gain since logic synthesis, allowing chip companies to focus their valuable engineering resources on differentiating their solutions.
- Faster simulation (either compiled, multicore, emulation, or FPGA-based prototyping) has been necessary to keep pace with Moore’s Law.
- Testbench languages led to new verification methodologies such as UVM and SystemVerilog, the latter of which is the foundation of modern design.
- System C and Transaction-Level Modeling (TLM) enabled virtual platforms, which have now gone mainstream. By enabling engineers to start software development months before hardware design is complete, virtual platforms are the industry’s best shot at getting software schedules under control and realizing revenue sooner.
In the absence of any one of these innovations, modern design couldn’t be accomplished, and our customers couldn’t complete a chip today.
EDA| How does Synopsys continue to foster innovation?
Constant innovation in every facet of the design flow is a pre-requisite to keeping pace with Moore’s Law. Whether our customers are designing mobile phones, automobiles, engine controllers, or medical devices, we must help them accelerate innovation so they can meet their aggressive time-to-market goals.
Innovation doesn’t happen in a vacuum. In EDA, it’s all about collaborating with our customers and ecosystem partners to help them solve their advanced design challenges. It’s important to be in early, to have the right partner(s), and to put very smart, dedicated people from different companies together to solve a seemingly unsolvable problem. Do this and you will get an innovative solution. As an EDA provider, it’s our job to be in the room so we can collaborate on co-invention and execution.
But that’s not enough. We have to hire the best engineers in the industry and give them room to innovate. It also takes investing massive amounts in R&D and integrating strategic Mergers and Acquisitions (M&As) to ensure that our customers have an EDA engineering partner that both understands the breadth of their challenges and has the horsepower to impact them. Synopsys spends more than 30 percent of our revenue on R&D – that’s a huge driver of innovation. In addition, we continue to use our strong cash position to engage in acquisitions that can accelerate the rate at which we address our customers’ problems and create solutions.
To truly foster innovation, however, all of this effort must be done in as open a fashion as possible so collaboration can extend across the industry. A great thing about EDA, more so than other industries, is that we have worked to maintain openness and interoperability among the tools in the flow. When taking this approach, technology success becomes dependent on pursuing the best idea, with every company motivated to out-innovate the competition in order to gain market share. The same level of innovation doesn’t happen in a non-interoperable flow.
EDA| We are hearing more about 3D ICs in the last few years. Do you think that true, stacked die, 3D ICs will require a change in design paradigm to create growth for EDA, or will most of the needs be fulfilled by existing 2D design tools?
Over the past few years, we have carefully looked at the technology and market landscape in close collaboration with our partners. We have concluded that the transition from 2D to 3D ICs is of an evolutionary nature; the vast majority of the 3D IC flow is identical to an advanced 2D design flow, and only a limited number of new capabilities and features are necessary to make EDA technology 3D-aware and capable. These enhanced tools are available now, and our partners are using them on real designs.
In the short term, 3D ICs will both complement transistor scaling to boost integration and performance to unprecedented levels, as well as represent a technically and economically viable alternative to scaling, extending the lifespan of established process technologies. Longer term, 3D technology will be a cornerstone of emerging innovative integration solutions, such as opto-electromechanical systems, and there are a great deal of opportunities for EDA to grow beyond the traditional IC level. As an example, multiphysics simulation and virtual, hybrid, and FPGA-based prototyping can play a fundamental role in the realization of highly heterogeneous systems, combining many flavors of ICs with MEMS and silicon photonics.
EDA| What is the biggest challenge facing the EDA industry today?
The EDA industry is healthy today because it is doing a good job meeting the challenges of modern design. To stay that way, however, we will need to continue to create innovations at an ever more rapid pace. New technologies such as 20 nm planar design and FinFETs introduce complexities that must be understood and abstracted away so designers can focus on what they want their design to do.
Every EDA company is motivated by the need to keep up with its customers and not disappoint them. Our customers are constantly inventing the next fundamental technology on the semiconductor roadmap, for example FinFETs. It’s up to EDA to help them do it by developing the fundamental software technology that enables them to design these advanced processes. They ask us to “invent a way to do that,” and EDA delivers. What if the semiconductor companies keep Moore’s Law going and we can’t? Then EDA will have let them down. As an industry, we have never let that happen and we never will.
A key challenge is having enough engineering bandwidth to solve the problems inherent in modern semiconductor design. Any individual problem is solved by the brilliant people who are working on it. The risk is that we end up with too many problems and not enough brilliant people to solve all of them. This is a key reason that Synopsys acquired Magma – to bring on board hundreds of excellent engineers to help deliver to the semiconductor roadmap.
Synopsys, Inc. | www.synopsys.com