• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Blogs
  2. Breakfast Bytes
  3. Anirudh's Keynote: A New Product...and an Acquisition
Paul McLellan
Paul McLellan

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
cadencelive 2020
cadencelive americas
Anirudh Devgan
cadencelive

Anirudh's Keynote: A New Product...and an Acquisition

21 Aug 2020 • 3 minute read

  Anirudh Devgan, Cadence's President, gave the keynote to open the second day of CadenceLIVE Americas. He titled it Computational Software for Intelligent System Design.

To avoid duplication, in this blog post I'm not going to talk about two aspects of verification that he covered in his keynote.

First, he announced Xcelium ML, our new machine-learning enhanced approach to random simulation-based verification. I covered that on the day he announced it in my post Xcelium ML: Black-Belt Verification Engineer in a Tool. I will also have a post in the next couple of weeks on Yoshi Watanabe's presentation later that day on how it works under the hood.

Second, he also talked about the infrastructure of verification across the whole product line (Xcelium simulation, JasperGold formal, Palladium Z1 emulation, Protium X1 prototyping). I covered that in my post Computational Logistics.

Computational Software

Anirudh talked about how computational software underlies much of what Cadence does. As he put it:

The key thing is that we want to leverage our core strength, over decades of experience in EDA, which is computational software, basically numerical, mathematical software. So if you look at the design part, the system part, and the intelligence part, the core vector is the computational aspect, or the numerical aspect of our software. So we try and take our core strength in computational software and of course supply to the design part for the best performance for our customers, and then extend that to system and intelligence.

Anirudh then moved on to verification, which I'm leaving out since I already covered it the week of CadenceLIVE Americas.

Digital Full Flow

Next up was digital implementation. We have fully integrated engines that talk to each other and deliver the best PPA (power, performance, area). We are in production deployment in 19 of the top 20 companies in the industry.

There's a lot we can do with digital full flow, and one key technology is, once again, the use of machine learning. Again, just like in verification, the engines themselves have to be best in class.
...
But during the course of the design, these tools are run over and over again, every day, or every other day. The designers use their intuition to get knowledge from one run and transfer that knowledge to the next run and optimize further.
...
Historically, there has been no way to transfer knowledge from one run to the next run. Actually, in all EDA tools, when you re-run the design the next day, it starts from scratch. There's no learning from the previous run. But using machine learning, it provides the mathematical framework to use the previous knowledge in the next run. This is similar to how a human would do it, so that is a very powerful shift, not only have a best-in-class flow for one run, but how to optimize the flow from one run to the next.
...
This is new. We are already working with some customers, but we would love to work with you.

We have a white paper if you want to dig deeper, How ML Enables Cadence Digital Tools to Deliver Better PPA.

Other Products

For the rest of his presentation, Anirudh took a whistle-stop tour of many other Cadence technologies:

  • OrbitIO, used for advanced packaging, 3DIC, system-in-package (SiP)—See my post System in Package? How to Plan and Build It
  • Clarity 3D Solver and Celsius Thermal Solver—See my post Under the Hood of Clarity and Celsius Solvers
  • RF Technologies—See my posts Designing Radios and Radar: AWR and The Integrand Story

Acquisition of InspectAR

The last thing Anirudh talked about was the acquisition of an AR (augmented reality) company called InspectAR. This allows you to combine EDA data with images of actual products. It's hard to explain something so visual in text, so I'll include a three-minute video so you get a better idea (and can see the motivation for the acquisition). There's more cool stuff on the InspectAR website. I will do a post on InspectAR in a couple of weeks, once I'm through with CadenceLIVE Americas (all over...bar some more writing), HOT CHIPS (which was earlier this week, so also all over bar the writing), and the TSMC Technology Symposium (next week). That should give you a good idea of what topics I'm likely to be covering for most of the next month.

 

Sign up for Sunday Brunch, the weekly Breakfast Bytes email