Real Intent: A sustained culture of Respect & Innovation
August 25th, 2016 by Peggy Aycinena
Real Intent, a Silicon Valley-based EDA company, has been underway since 1999. That’s a lot of time for a company to continue to succeed amidst the shifting sands of an industry that specializes in either acquiring smaller companies or under-cutting their prices until the smaller companies simply close their doors.
In other words, Real Intent is a survivor and tells a remarkable story of steady perseverance and a well-established, respectful relationship with its customer base. Without both of these things, the company could not have remained viable.
This week, I enjoyed an excellent phone call with company CEO, Prakash Narain, a conversation of particular interest because Dr. Narain spoke candidly of the challenges that face small EDA companies in the current business climate.
Narain said, “Yes, Real Intent has been around for quite a while. We started out as a company for formal verification, but I believe we started too early. While the company made some innovations and advancements [in the technology], we eventually grew tired of waiting for the formal space to evolve, something that finally took place 10 years later.
“Hence, we moved in the space of static sign-off. It is our belief that the most under-utilized [strategy] and biggest opportunity for improving the verification flow in terms of design is static sign-off.”
Responding to a request for clarification, he continued, “Before defining static sign off, first let’s understand verification.
“Verification is the process of having a collection of failures that you want to make sure are not going to happen, and looking for them in your design. When you have fixed those [potential] failures, you know they won’t happen. And this is typically one paradigm of thinking about verification.
“The other paradigm, which is used in simulation and emulation, is basically as follows. We are going to compare two models of the design and find any mis-matches between the two models. This paradigm for verification, however, is finding less and less application today.”
Why has this strategy of comparing two models of a design fallen out of favor?
“That was a good paradigm,” Narain replied, “when designs were happening from scratch.
“According to some estimates, in any new SoC today the number of new gates is at 40 percent or even less. The rest of the design consists of reused gates. As a result, the designs have become very massive and an untargeted approach to verification [is no longer viable]. Verification has become much more targeted.”
“Please define targeted,” I asked.
Narain said, “You look for error symptoms, and you look to be sure those symptoms are not present [in the design].”
“But how can you know that you’ve spotted all the symptoms?” I asked. “How do you know when your verification is done?”
Narain said the question has two parts: “First, are you sure you are checking for everything, and then have you checked enough?
“This second piece is the limitation of simulation – you do not know if you’ve checked enough. Then the question typically is, how long do you keep simulating? And it’s this second question that’s adequately answered by static sign-off.
“Which problem am I checking for? Once that’s determined, the whole process is built to statically analyze the design, or the existence of that particularly problem, assuring therefore that you are done.”
“So if you finish with static sign-off,” I asked, “there’s some high degree of confidence that everything’s been anticipated?”
“Yes,” Narain answered. “It can be used in problem areas until ultimately, they get totally clean sign-off and people then have a high degree of confidence. We don’t ask the question of when you’ve done enough static-timing analysis. There is enough verification [using this strategy].
“Consider also cross-domain clocking verification. Again, it cannot be covered in simulation, only in static-timing analysis. People don’t ask when there’s been enough CDC sign-off, because all failure modes have been examined and verified.”
“There are then no problems?” I asked.
“Exactly,” Narain said. “Static timing analysis requires specific development of technologies, specific to each failure mode – clock-domain crossing, or stating timing, or for that matter, some functional mode.
“It is a collection of applications. What is required is that a particular domain is picked up, and wherever the custom design is tenable to that application.”
Narain expressed optimism: “We are starting to see [whole new] areas where these [strategies] apply. We believe we have just scratched the surface of the opportunities here, that this technology will continue to grow and add significant value to verification.”
“Nonetheless,” I persevered, “how do we know we have addressed all of the domains?”
“Will we set an absolute bar, you are asking,” Narain replied. “Or are we looking for improvements in specific processes. Both, because increasingly we will begin to rely on static sign-off. Whatever the future, that has to be the conclusion.
Narain invoked the past: “If you look at the history of verification. We have advanced gate-level simulation, yet timing simulation is almost never used.
“Go back 15 or 20 years, these may have been important steps in the design process, but they lost their place [of importance] because the requirement is no longer there.”
“Do your customers also ask questions about the viability of static sign-off?” I asked
Narain said, “First, we are talking about fundamental innovations here, and incremental paradigm shifts. Our customers are basically looking at our products, and appreciating our offerings for the applications – those things they are looking to deploy in their design flows.
“Our customers are also articulating their existing problems. For example, CDC customers are also alerting us to the problem of SDC constraints. We examine solutions and work with the customers [across] multiple applications.”
“Static sign-off is about inventing new technologies,” he continued, “it’s dramatically about product design, usability and product design.
“We can invent the technology, cut cannot do full product design without customers’ feedback, how it fits into the work flow, to really understand the requirements.
“To enable static sign-off, you need innovation and you also need product design innovation where can you accurately conceive the [manner in which the product is used].”
“You would need a set of customers who trusted you enough to be candid about their problems,” I commented.
“Absolutely,” Narain confirmed, “and that trust comes out of being there for the customer, and being able to deliver on commitment, innovation, and a road map.
“Building trust is a very complex process, but the customers who have developed that trust in us have gained very significant benefits – and us from them.”
“Do you think that kind of trust is related to the longevity of the company?” I asked.
Narain said, “Longevity is not the right requirement. It’s the culture [that matters], a culture of partnership with customers. A willingness to understand the problem, work on the problem, and thereby support the customer.
“And it doesn’t matter what type of customer – an FPGA company, an application processor company, across the board. It’s a culture and willingness on the part of the customer to recognize and reward the innovation [we bring to the work], and along the way to enjoy the benefits [of our collaboration].
“This is how we are striving to drive excellence in the product. This excellence doesn’t happen quickly. It take times, but longevity by itself is not enough. It requires a culture and relationships with customers to enable them to take advantage of our innovations.”
“Can you look ahead five years?” I asked. “Will there be more paradigm shifts?”
“It’s difficult to look ahead five years,” Narain replied. “Innovation is organic, we do not know where the next great idea is going to come from.
“Our great ideas here at Real Intent have come from different people under different circumstances. Someone sees a good idea and runs with it, that’s the formula we understand. But we don’t know where the next great idea will come from, we cannot see a pattern.
“Having said that, in five years our biggest innovation will be more great contributions to verification technology. We are that excited about who we are and what we are doing.”
Our conversation turned to internal CAD tool development versus relying on third-party tool vendors; I referenced Narain’s own background at both IBM and AMD in those areas.
He said, “Yes, it used to be a slam dunk that the internally developed tools were preferred. The idea was to innovate by building unique CAD solutions within the company, so you didn’t give the world advance knowledge of your design problems.
“But over time, [the industry] has learned, investing in internal innovation means you become insulated from innovation that is happening elsewhere, outside of the company. The other negative – you also become a prisoner of your own CAD flows and methodologies.
“Once the advantage of that internal innovation is lost, it’s harder and harder to advance the internal solution. These systems have to be built, and then modified and changed as demands evolve. Eventually the pace of change in design and the costs [of continued innovation] become prohibitive.
“An EDA company, however, can provide a design solution that id generalized and changeable. Even those customers who believe in their internal innovation, find that through partnering with vendors they can move faster, their insights into design methodologies [gained by working with EDA vendors] are improved.
“As tool vendors, basically we are enabling the customers who are running fast – of course, we respect their confidentially – and they benefit from our rapid innovation [in response to their problems].”
Admiring Narain’s enthusiasm, I asked about exit strategies for Real Intent and referenced recent EDA acquisitions.
He responded, “static sign-off is not like formal – we are not Jasper. But in reality, there are three public companies in EDA and yes, exits happen in this industry through acquisition.
“Real Intent can continue, however, because we have access to funding. Yes, our investors at some point will look for return on their investment. It’s not mysterious here in EDA – a typical exit is an acquisition, but it takes two to tango. And when it happens, it requires the possibility and a willingness.”
“Is staying independent a challenge in the current environment?” I asked.
Narain was candid: “There are three problems. First, insufficient respect from the customer companies for the innovation that the EDA companies attempt to provide.
“Second, the pricing and policies utilized by the large EDA companies to simply squeeze the smaller companies is an ongoing problem.
“And third, EDA is not attracting a lot of investment currently, and without investment nothing happens. So you’re seeing fewer small companies, less start-up activity and, as a result, I believe the pace of innovation has slowed.”
“I do not deny,” he added, “the attempts by the large companies to innovate, and I have a lot of respect for those efforts, but the sheer number of innovations are small. If we had twenty or thirty or forty companies in EDA, many more ideas would come about more quickly.”
“Perhaps 5 years from now,” I suggested, “this endless cycle of consolidation on both the customer and EDA side will reverse. Perhaps we will once again see healthy chaos as entrenched interests break apart and the innovation cycle starts anew.”
Narain agreed: “Yes, certainly I will make a note to that effect. In 5 years, perhaps we will be talking about a renewed energy and opportunity for innovation within EDA.”
In closing, I asked, “Do you enjoy your work?”
“At the end of every day I feel extremely tired,” Narain chuckled, “but then I very much look forward to the challenges of the next day, and the next year!”
Real Intent CEO …
Dr. Prakash Narain’s career spans IBM, AMD and Sun where he got hands on experience with all aspects of IC design, CAD tools design and methodology. He was the project leader for test and verification for UltraSPARC IIi at Sun Microsystems. He was an architect of the Mercury Design System at AMD He has architected and developed CAD tools for test and verification for IBM EDA.
Dr. Narain has a Ph.D. from the University of Illinois at Champaign-Urbana where his thesis focus was on algorithms for high level testing and verification.
Tags: EDA, Prakash Narain, Real Intent, Static sign-off