Hanover, Germany, Feb. 27, 2017 – Dream Chip Technologies announced today the presentation of the industry`s first 22nm FD-SOI silicon for a new ADAS System-on-Chip (SoC) for automotive computer vision applications at the Mobile World Congress in Barcelona. The SoC was created in close cooperation with ARM, Arteris, Cadence, GLOBALFOUNDRIES, and INVECAS as part of the European Commission’s ENIAC THINGS2DO reference development platform.
The SoC offers high performance image acquisition and processing capabilities and supports convolutional neural network (CNN) vision workloads to meet the demand for complex automotive object detection and processing. This ADAS SoC is highly optimized for high-end computer vision performance at a very low-power consumption to enable autonomous driving in production and is capable to support ADAS functions like road-sign recognition, lane departure warning, driver distraction warning, blind spot detection, surround vision, park assist, pedestrian detection, cruise control and emergency braking.”
The design incorporates Dream Chip Technologies’ image signal processing pipeline in conjunction with Cadence Tensilica Vision P6 DSPs and a quad-core cluster of ARM® Cortex®-A53 processors. In addition, a lock-step pair of Cortex-R5 processors provides ISO 26262 compliant functional safety and the SoC is interconnected with an Arteris FlexNoC network-on-chip. The SoC uses multiple IPs such as foundation IPs, LPDDR4, PLL, Thermal Sensor and Process Monitor, from INVECAS. Cadence’s LPDDR4 controller and INVECAS’ LPDDR4 PHY IP provide two LPDDR4 3200 high bandwidth memory interfaces.
Dr. Jens Benndorf, Managing Director and Co-Founder of DCT said: “It was a unique experience to create and coordinate such a powerful team and successfully lead the consortium to two tape-outs. In this project, we have delivered silicon using a brand new process technology for the automotive industry.” He adds: “The project not only shows the strength of the European semiconductor industry, but also an ability to collaborate efficiently to provide technology needed urgently by the industry to power advanced automated driving solutions.
Dream Chip Technologies has designed many highly complex SoCs for customers worldwide and was selected as the design service lead for this THINGS2DO project. The project has created a camera-based ADAS reference platform which benefits automotive companies through advanced technology and by shortening design cycles and time-to-market for automotive innovation. The SoC is fabricated on GF’s 22FDX® semiconductor process at the foundry’s Fab 1 facility in Dresden, Germany.
The new ADAS platform is targeted at automotive Tier-1s with a need for cost, performance and power-optimized SoCs for a range of ADAS applications and with potential for customization.
The choice of a quad-core Cortex-A53 processor configuration is a popular choice for computer vision applications to manage automotive vision applications. More advanced autonomous driving systems can also be enabled with this level of computer vision capability, including those using multiple smart cameras.
“This project provides a boost for European chip designers focused on solving some of the automotive world’s most complex problems,” said Nandan Nayampally, general manager, CPU Group, ARM. “By using a quad-core ARM Cortex-A53 configuration, with the Cortex-R5 fortifying the safety critical aspects, Dream Chip has produced a highly efficient and functionally safe processing solution that sets a high bar in the automotive vision sector. It underlines the rapidly increasingly deployment of ARM technology in vehicles; ranging from detection sensors to infotainment, ADAS and autonomous driving.”
“ADAS features are exhibiting the highest growth within the automotive electronics domain as vehicle makers look to differentiate on enhanced products that provide real-time vision processing,” said Rajeev Rajan, vice president of IoT and Automotive at GF. “GF’s 22FDX is an ideal platform to design power- and performance-efficient solutions for automotive vision and camera ADAS applications. As an industry first, this milestone affirms our commitment to support and collaboratively shape automotive markets.”
Dasaradha Gude, CEO of INVECAS said: “We are excited to contribute to this program as this collaboration further strengthens INVECAS’ position as a key provider of silicon proven IPs for GLOBALFOUNDRIES’ 22FDX technology. INVECAS’ rich family of IP offerings ranging from Foundation IP, Interface IP like Multiprotocol SerDes, DDR34/LPDDR34, MIPI, HDMI etc. enables complex SoC designs for the automotive market addressing the next wave of connected car opportunities in Europe. Our objective is to provide silicon-proven IP and ASIC solutions to address the challenges of ever growing design complexity faced by the semiconductor industry today.”