Edge AI

Huge advances in electronics have been made by giving devices access to the computing and storage power of sophisticated server farms via the internet to provide a high level of Artificial Intelligence (AI). This is known as cloud-based processing. But these farms use a massive amount of electricity giving them a large carbon footprint let alone all the power used by the internet infrastructures.

In the drive to reduce carbon emission, we are designing solutions for customers that have the AI in the device – known as Edge AI. It’s our sweet spot of designing ASICs that integrate a large number of features into one chip.

Edge AI refers to the deployment of AI technologies directly on edge devices, such as smartphones, IoT devices, cameras, drones and other embedded systems, instead of relying on cloud-based processing. The approach offers several benefits, including reduced latency, improved privacy and security, and increased efficiency as well as cutting out the massive amount of power required for cloud-processing.

Here are some typical applications of AI at the edge:

• Autonomous Vehicles
• Smart Surveillance
• Industrial Internet of Things (IIoT)
• Healthcare Monitoring
• Edge Robotics

Automotive

We are a major supplier of chip designs for Automotive applications because we understand the complex, highly demanding standards of ISO 26262 and Functional Safety (FuSa). Key to this are our Architecting the Future architectural frameworks from which we craft custom ASIC designs to meet exacting technical and commercial requirements. We have ADAS-specific, architectural frameworks designed for fast-tracking ISO 26262 Functional Safety compliance.

Automotive chips play a crucial role in modern vehicles by providing computing power and enabling various functions and features such as sensor control.

Here are some typical applications of automotive chips:

• ADAS 
• In-Vehicle Infotainment Systems
• Telematics
• Engine Control Units
• Electric Vehicle Systems
• Advanced Lighting Systems
• LIDAR & Cameras

8K Video

Video has been a core strength of Sondrel for many years and we have designed numerous ASICs for customers across a wide range of application areas from surveillance to games consoles, building up a bank of expertise that makes us one of the leaders in this field. When it comes to handling big data streams, we're the experts.

An example of recent work was for a multi-spectral computer vision technology specialist. We created an ASIC for facial and object recognition from video streams on end node, security camera devices. This 400 mm² chip incorporated two quadcore multithread CPUs and was on a 16nm node process.

Here are some typical applications for 8K video:

• AR/VR
• Medical Imaging
• Architecture and Design
• Surveillance

Smart Homes

When it comes to turning a myriad of single chip solutions into one ASIC to reduce costs, size and power consumption, we are the first choice for companies working in this area. We know how to integrate new features such as multiple sensors to make their devices smarter and give them that vital competitive edge over rivals. When you need a complex digital ASIC, we are your solution to create smarter devices.

Smart homes utilize Internet of Things (IoT) devices and automation technology to enhance comfort, convenience, security, and energy efficiency within residential spaces for an enhanced user experience. Here are some typical applications of smart home technology:

• Home security
• Lighting Control
• Energy Management
• Home Entertainment
• Automated Appliances
• Digital Assistants

Wearables

Sondrel’s key strengths include designing for low power operation and for minimal area. Laying out all the IP blocks needed for an ASIC is relatively easy at the first pass – it’s rather like putting all the pieces of a jigsaw out on a table top. The skill is connecting them all together in the right locations to optimise the flow of data between the blocks without bottlenecks and areas of overheating. And the area has to be minimised to keep costs down though this is partially helped (or complicated) by having multiple layers for interconnections making this a three-dimensional challenge. Or four dimensional when you consider timing issues.

The ideal solution inside a smartwatch is all the electronics on just one chip rather than numerous chips as they take up much more area -- the sum of the area of several ICs being greater than an all-in-one. There are few companies with the skills to design such a complex, highly integrated ASIC which is why we are in discussions with a couple of companies in this area.

Here are some typical applications for wearables:

• Fitness and Health Tracking
• Heart Rate Monitoring
• Sleep Monitoring
• Notifications and Communications
• Music and Media Control
• Hearing Aids

Consumer Devices

We have designed SoCs for dozens of innovative consumer devices over the years from advanced game consoles to phones. We’ve fitted them with Bluetooth, WiFi, all manner of sensors and ever more powerful processors because the next generation of a product has to be smarter than the previous one to ensure continued product sales and differentiation from rival offerings.

We work closely with companies to work out what new features can be incorporated and still meet the constraints of end product costs. Essentially working forwards and backwards to determine a solution that balances features, performance and costs. It’s always a compromise but we bring to the discussions our intimate knowledge of the costs of designing and making chips so that the best possible balance can be achieved for Power, Performance and Area. And bring new ideas to the table to enhance a project based on innovations we have created for other customers plus new ones just for you.

Here are some typical applications for consumer devices:

• Smartphones
• Laptops and Computers
• Gaming Consoles
• Smart TV’s
• Digital Cameras