MENU

Computational Software Q&A with Frank Schirrmeister, Cadence

Computational Software Q&A with Frank Schirrmeister, Cadence

News |
By Ally Winning



What are the key trends influencing electronic design automation (EDA)?

The areas of EDA and what, in the financial world, is often referred to as technical software, including simulation and analysis, mechanical CAD and product lifecycle management (PLM), are really interacting more closely and merging as the design chain of OEMs—developing electronic systems—and semiconductor providers undergoes significant transformations. As to the markets, in 2019, the EDA and semiconductor IP market of $10B+ served as an enabler for the $430B semiconductor market (according to Omdia Research) that in exchange fuelled a $2.4 trillion OEM electronics market. By 2024, the semiconductor and OEM markets are expected to grow to roughly $530B and $2.8T respectively.

What are some of the vertical consumer end markets that drive all of this?

End-market demand for electronics in eight key verticals—consumer, compute, mobile, networking, automotive, aerospace/defence, industrial and healthcare—is very strong. Some key subsegments stand out due to their semiconductor growth potential, namely artificial intelligence/machine learning (AI/ML)-enabled smart assistants, game consoles, augmented reality/virtual reality (AR/VR) in the consumer vertical, data centre servers and storage in compute, mobile communications infrastructure and security in networking, and ADAS, hybrid and connectivity in automotive, as well as automation in home and industry, and even virtualization in healthcare and aerospace/defence.

You mentioned changing design chains. Can you please elaborate?

The dynamics within the industry design chains are poised for change, with OEMs in automotive, hyperscale companies in compute and new players for next-generation networking simply reshuffling the design chain—just like the mobile industry did during the past two decades. Several of these industries are interconnected, with IoT sensors and terabytes of data collected in cars linked via next-generation networks for compute at the edge and in data centres, with data being the universal fuel. For instance, traditional system companies like Facebook, Google, Amazon and Microsoft, often referred to as the Tier 1 “hyperscalers”, have been influencing electronic design with their contributions to initiatives like the Open Compute Project. As a next step—transforming the industry—they are increasingly designing their own silicon as well. Think the Amazon AWS Graviton instances in the cloud based on Arm processor technology at lower power points.

What does all of this mean for EDA and technical software?

The uptick in the EDA and semiconductor IP space opens new growth areas for these classic domains, which encompass digital and custom implementation and verification and are now expanding towards system design and eventual pervasive intelligence. We call this expansion, “Intelligent System Design”. The fundamental technology shift here is the broader application of computational software. Electronics and electronic environments, including electromechanical, power and thermal aspects, need complex computation and simulation to efficiently apply trends for parallelization and hardware-assisted execution using and enabling implementation of domain-specific architectures. All of these considerations require cross-functional technology aspects like safety, security, low power design and advanced-node implementation at 3nm, as well as new assembly aspects like 3D-IC and chiplets.

Can you provide some examples of computational software?

Computational software supports and manages the complexity of fundamental industry trends—hyperscale computing, 5G, AI, industrial IoT (IIoT) and autonomous driving—and transforms electronics across the verticals we discussed earlier. For instance, finite element analysis (FEA) is widely used for structural analysis, calculation of heat transfer, fluid flow and mass transport as well as electromagnetic analysis, which is needed extensively when assessing silicon-in-package (SiP) effects. It all goes back to complex math, like partial differential equations in two or three space variables. FEA uses a numerical method for solving them by subdividing the larger system into smaller finite elements using space discretization via a mesh of the object that is to be analyzed. This is then boiled down to a system of algebraic equations, and FEA approximates the unknown function. Other examples of computational software are the solvers used in formal verification, as well as advanced algorithms for constraint-solving as used in verification. Technologies like SPICE predict the timing, frequency, voltage, current or power at the transistor level. Heuristics like randomization strategies, simulations, inference engines, graph methods, probability estimation and sampling are used in AI, and pattern recognition algorithms like classification, regression, macro modelling and optimization trees are used in silicon layout.

Why do you see EDA taking centre stage here?

Computational software has been at the core of the EDA industry for several decades, delivering major capability and productivity advances in hardware design and enablement of software development. Given the complexity of chip and system design, the application of highly tuned algorithms across diverse applications that can operate on enormous data volumes and the computational aspects have, in recent years, been split to allow massively parallel execution. Now we witness the close dependencies, if not the merging of software and hardware development, as well as the electrical and mechanical aspects. Time-to-market windows must be met, so early and continuous integration, basically integrating and simulating as much as possible of electronic products before its production, becomes a fundamental requirement.

What’s in it for the end users—you and me in our everyday lives?

We get higher quality products on time, and a user experience that is top-notch. Take hyperscale computing, for instance. A lot of us have fitness and health trackers. They collect data and deliver it via our phones. Some aspects, like sleep analysis, are either processed locally at the endpoint or edge, or, if more complex, are sent using LTE and future 5F networks to local edge-compute nodes. More complex analysis is done in the actual hyperscale data center. Which data is stored at which location and where processing takes place in the network, highly depends on the required latency that the user expects for results to be available. All of it can be simulated using early and continuous integration, as well as computational software to assess issues like thermal analysis and electromagnetic effects.

Bottom line, classic EDA is merging with the world of system design and AI. It’s an exciting time for EDA and IP, where these key technologies will serve as the enablers of a bright future for electronics consumers.

More information

www.cadence.com

Related news

Tensilica Vision DSP doubles vision and AI performance

Cadence releases functional verification IP for USB4

Cadence JasperGold helps Hitachi to comply with IEC 61508 Series SIL 4 requirements

Cadence and Nvidia partner to apply machine learning to EDA

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News

Share:

Linked Articles
10s