Automotive: CFD, FuSa, Aging, Vision, Light

16 Feb 2022 • 5 minute read

That's a lot of acronyms and buzzwords! Just in case you don't know them all, they stand for:

CFD: Computational Fluid Dynamics
FuSa: Functional Safety
Aging: Not just getting old, but in the context of automotive it means analyzing transistor performance (especially for analog) over the twenty-year lifetime of the chips.
Vision: Well, you know what vision is, George H.W. Bush famously "didn't do the vision thing". But Tensilica has a whole vision product line heavily used in automotive (among many applications).
Light: You know what light is (although probably don't really know: is it a wave or a particle?), but in this case Light is a Silicon-Valley-based company producing chips for automotive applications.

Electronics in Cars

Let me tell you about Cadence's system for designing tires. Well...of course, we have no system for designing tires. But tires are increasingly tied up with electronics. My previous car had no tire pressure monitoring. My current car (a Mini) has enough monitoring to tell me my tire pressure is low in a particular tire. I was at the dealership recently, and in the courtesy car bringing me home I noticed the car now has the actual tire pressure in all 4 tires on the dashboard. Inside the tire there is a pressure monitor and it communicates wirelessly with the car to report. So even something as prosaic as tires has an electronic component. By the way, I have no idea if Cadence was involved in any of these tire electronics, I'm just pointing out the trend. Even the most basic parts of cars are increasingly electronified, if that's a word (it isn't).

This post is mostly going to be a summary of each of these features of the automotive ecosystem, with pointers to previous posts I've written on the topics.

CFD

The old way to design cars was by making accurate models and then testing and making measurements in wind tunnels. This is very expensive. Computational fluid dynamics can be used as a sort of virtual wind tunnel, to make the measurements of things like wind resistance in the computer. CFD is also used in engine design to analyze thermal issues, design turbines, and so on.

Here is a post about a recent conference: Highlights Of The Automotive Aerodynamics And Thermal Management Conference. One quote from that post to give you an idea of the scale of resources required:

Audi developed the recent Q4 eTron using 5 million CPU hours and 250 Wind Tunnel hours over 40 months, focused entirely on aerodynamic and aero acoustic optimization.

And this page, End-To-End Aerodynamics CFD Simulation of a Car, has a video of the webinar (registration required). Summary paragraph:

Optimizing the external aerodynamics of a car can greatly improve vehicle performance and reduce fuel consumption. It can also significantly enhance passenger comfort by reducing cabin noise. And accurate prediction of front-end airflow enables optimizing thermal management of key components like heat exchanger, disc brakes, etc..

Here is a video in the Designed with Cadence series by Honda:

FuSa

FuSa stands for functional safety. I still like Alessandra Nardi's pithy one sentence summary, which I discuss in my post Make Sure Your Car Doesn't Break Too Often...When It Does, Make Sure You Catch It.

Cadence recently announced our solution in the space, see my post Announcing Cadence Safety Solution and the Midas Platform...Turn Your Automotive Products into Gold. Summary paragraph:

Today, at CadenceLIVE Europe, we announced the Cadence Safety Solution, a new offering targeting safety-critical applications and featuring integrated analog and digital safety flows and engines for faster ISO 26262 and IEC 61508 certification. The solution, which includes a new FMEDA (Failure Modes, Effects, and Diagnostic Analysis) offering called the Midas Safety Platform, allows customers to perform FMEDA-driven analog and digital verification of safety-critical semiconductors for advanced automotive, industrial, and aerospace applications. The Midas Safety Platform also provides a GUI-driven use model to do an early architectural FMEDA analysis. Once RTL/gate-level design data becomes available, the Midas Safety Platform can carry out a detailed FMEDA analysis that is more accurate.

Also see my post Accellera Functional Safety. This is mostly about the standardization efforts for functional safety. As it happens, the aforementioned Alessandra Nardi is chair of the working group.

Aging

Transistors in integrated circuits wear out over time. This has always been true but in the distant past, problems would not arise until long beyond the lifetime of the chip, but in more advanced node processes, this is a bigger problem. The other change is that in applications like automotive, the chips need to last for a couple of decades or more without any issues. Although this process is called aging, it is not an effect of time but rather how much activity the transistor experiences. It is mostly a problem in analog designs, since there is a lot of margin in digital designs.

Analysis is based on something called, for obvious reasons, a bathtub curve. In early days there are potentially early failures and they need to be weeded out before the chips get into the cars. At the other end of the lifetime, aging effects can lead to failures.

For more details see my posts:

Vision

Modern cars need to see. Of course, autonomous driving will require a lot of capability, but even advanced driver assistance system (ADAS) features like lane following or automatic emergency braking (AEB) require cameras. Cadence has a whole family of Tensilica vision processors, the most powerful of which is the Vision Q7. See my post Vision Q7 DSP: Real-Time Vision and AI at the Edge. These vision processors do not just do the actual vision, they also have neural network capability for inferences (recognizing where the lane markings are, for example).

Light

One company that has been using the Tensilica Vision Q7 is Light. As announced in the recent press release Tensilica Vision Q7 DSP enables Light’s Clarity Depth Perception Platform to power next-generation ADAS systems with 10X greater performance than quad-core CPU.

In a bit more detail:

Light’s solution enables machines to see better than humans by using two or more cameras, novel calibration and unique signal processing to provide unprecedented depth quality across the camera’s field of view. Integrating the Tensilica Vision Q7 DSP into Light’s solution enables up to 10X faster processing of measured depth compared to a quad-core CPU, improving real-time accuracy and reliability. The highly configurable and extensible Vision Q7 DSP is accompanied by an array of optimized computer vision libraries, optional accelerators and toolchains that efficiently handle such workloads. Light implemented its proprietary multi-view depth perception algorithms via custom instructions using the Tensilica Instruction Extension (TIE) language, resulting in an additional 4X performance improvement with 3X area/power savings.

Learn More

Follow any of the links embedded in this posts that points to my previous blog posts on each topic.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.