On September 17th, 2017, Ian Dennison (Sr Group Director R&D, CAS Product Marketing, Cadence) presented a fascinating keynote on industry trends, challenges, and opportunities in electronic design. His keynote addressed the disruption and opportunities in the automotive vertical market facing the EDA industry in the coming years. Agility, Change, and New Design Fabrics within the Automotive Ecosystem We’re poised for a huge change in the transportation industry. Are you ready? Check out this slide, which opened Ian Dennison’s presentation on Disruption and Opportunity in the automotive field in the coming years at CDNLive Korea. While we are on the path to autonomous vehicles (he called it AV, I call it Auto 2 —that is, Autonomous Automobiles), we’re coming up on challenges to the development cycle, new entrants into the field (why, hello, Tesla), higher compute rates, and a whole host of other challenges. The EDA field is also perfectly poised to address these problems. With Cadence’s new strategy of System Design Enablement—that is, ensuring that the entire system design is considered from the top-down, and each component of the system can be designed and developed in tandem—the route to Auto 2 is relatively straightforward, despite its complexity, particularly regarding PCB design. Looking at safety, the ADAS system takes a boggling amount of parallel design. For example, combining the electronic and photonic design of chips and then boards require a combination of Lumerical System Simulation software, Spectre Electrical Simulation, the Virtuoso Custom Design Platform for analog design (for schematics and other forms of layout), and PhoeniX Photonic layout generation just to make the components that are then shipped to a fab shop. It certainly helps that these components work seamlessly together. Voilà, you know that it will work, and in the timeframe required. The ADAS system uses lidar to sense the environment around the car, with layers upon layers of complexity. To design this system, you can use Allegro layout tools, Spectre RF signal simulation, Allegro Sigrity, MathWorks tools for beamforming, calibration, and enforcing wireless standards, not to mention using the Virtuoso design platform. Again, all must work together, seamlessly. The Cadence System Design Enablement Strategy comes to the rescue. Layout-dependent effects at every level regularly confound designers, sometimes changing transistor characteristics by as much as 80%. How can anyone work that way? Using Virtuoso Advanced Node, all characteristics are taken into consideration from the very beginning, and the Virtuoso ADE Product suite allows for real-time tuning, which virtually eliminates the endless design loop. Now, looking at CNNs, for a car to detect “pedestrian” or “light pole” or “motorcycle”, a single image can take as much as 10 12 MACs. With newfangled moving pictures (aka video feeds), multiply that by the number of frames per second—and you have a whole bunch of data that must be processed as fast as a car accident. The Tensilica C5 DSP for CNNs meets that challenge, with a throughput of 1 Trillion MACs/second/1mm 2 silicon area. Nothing is faster when processing neural network data. The C5 can be combined with Vision P5 and P6 DSPs, as well, and uses the same toolset. The Tensilica C5 DSP has the highest scoring German Traffic Sign Recognition Benchmark (GTSRB) algorithm in the industry. From the workstation simulation to RTL Simulation, to Emulation, to FPGA prototype to the first silicon, Cadence offers tools at every stage of design, development, and signoff. At Cadence, we call it System Design Enablement. In the automotive vertical market? We call it… Automotive Design Enablement. Ian Dennison closed the presentation saying that disruption and opportunity in the automotive market will obviously create winners and losers. The winners are the ones who use the latest design fabrics, and that includes adopting the Cadence Automotive Design Enablement strategy. Go forth and make the opportunity happen!![]()
↧