Latest Automotive and Industrial News from Lattice
Posted 07/31/2019 by Lattice Semiconductor
Posted in
  • Consumer
Bring More Performance to the Edge

Need to build higher performance AI solutions for the edge? Lattice's sensAI is ready to support your next project. Lattice has recently announced major performance and design flow enhancements to its new sensAI solutions stack. Suppling all the key hardware and software components designers need to implement low power, always-on, FPGA-based Machine Learning/AI functionality in smart edge devices, the latest rendition of the stack offers a 10X performance improvement over its predecessor. Now developers can take advantage of a seamless interface with support for new neural network models, more machine learning frameworks and faster design cycles. At the same time new customizable reference designs help speed development of applications like object counting and presence detection. And a rapidly expanding partner ecosystem offers developers new custom design services that accelerate development.

New Radiant IP & Reference Design

To make development and debugging with iCE40UltraPlus easier, Lattice rolled out Radiant Software, which provides basic building block IP functions including PLL, DSP, Memory, I2C and SPI. In addition, reference designs that allow users to take advantage of Radiant include:

  • Graphics Acceleration
  • Human Presence Detection
  • Image Sensor Bridge
  • PDM Microphone Aggregation
  • Sensor Data Buffer
  • Sensor Interfacing and Preprocessing
  • Simple Sigma-Delta ADC
  • Single Wire Aggregation
  • SPI Slave to PWM Generation
  • SPI to MIPI D-PHY
Demos Showcase sensAI Capabilities, Ease-of-Use

Interested in sensAI, but unsure you have the time and energy to invest in learning a new technology? Check out Lattice's technology demos. At the recent Embedded Vision Summit the Embedded Vision Alliance created videos of a number of new demos of sensAI stack across multiple applications. One demo illustrates how easily a Lattice ECP5 FPGA running sensAI can quickly detect and classify 100s of classes, turning appliances into smart IoT devices in homes. In the second a Lattice iCE40 Ultra Plus FPGA running the sensAI stack can perform simple gesture detection tasks while operating at < 3.3 mA when active. Yet it still delivers enough performance to enable smart vision without having to connect to the cloud for data analysis. Both of the above-mentioned demos are available on the Embedded Vision Alliance's YouTube channel.

Industry Awards Keep Piling Up for sensAI

If everyone likes a winner, the electronics industry must be falling in love with Lattice's sensAI solutions stack. The solutions stack has racked up five prestigious industry awards in just the last six months. The wins started back in January when sensAI was named an 'EDN Magazine Hot 100' product for 2018. By combining modular FPGA hardware kits, neural network accelerator cores, software tools, reference designs, and custom design services, Lattice's sensAI stack helps fast track development of AI inferencing in edge devices for smart homes, smart cars, smart factories, smart factory applications.

A few months later in April, sensAI was named 'Most Competitive FPGA Product' by China Electronics Market (CEM) as part of its 2018 Editor's Choice Awards. Among its many capabilities sensAI was chosen for its ability to combine ultra-low power consumption, small package size, and flexible interface IP support into a compelling solution for adding AI inferencing capabilities to the millions of power and form factor sensitive devices operating at the network edge.

In late June Lattice's sensAI solutions was designated the 'Best AI-based Solution for Engineering' in the 2019 AI Breakthrough Awards, a forum that recognizes the top companies, technologies, products and services in the field of AI. The award typically attracts up to 1500 nominations from all over the world.

A week later sensAI did it again when it was named Internet of Things 'Product of the Year' at the Electronics Industry Awards held in London. 'The full featured sensAI stack includes everything you need to evaluate, develop and deploy FPGA-based ML/AI solutions - modular hardware platforms, example demonstrations, reference designs , neural network IP cores, software tools for development, and custom design services,' said Niamh Marriott, Components Editor for Electronics Magazine. 'Despite tough competition this product manages to transcend the pack.'

In early July sensAI racked up its fifth industry win when it was tapped for the Assodel award for the industrial and automotive-related hardware product category. The Assodel recognizes the best manufacturers in the electronics sector and is a part of the Italy Electronic Districts Association.

Press Praises MachXO3D and sensAI Announcements

Lattice's recent announcements of the MachXO3D FPGA and a major upgrade to its sensAI solutions stack have resonated well in the industry press. Patrick Moorhead, writing in Forbes Magazine, recently said 'I believe Lattice's latest, low power FPGA offerings stand out in the crowded, high power field for edge AI and security. I am very interested to see many of the end products come out,' he added.

Richard Nass, Executive VP for OpenSystems Media, added more praise for the recent upgrade of sensAI in Embedded Computing Design. 'Lattice is already known for delivery of low power programmable devices and continues that trend with this release while embarking down the AI path,' he noted.

Finally, Kevin Morris of Electronic Engineering Journal chipped in with a similar opinion. 'The company (Lattice) is hitting at two of the most critical aspects of connected systems these days - securing applications and embedding AI inferencing at the edge - and they have a decent head start on what is definitely becoming a highly competitive segment.'

Tags:
  • iCE40 UltraPlus
  • ECP5
  • sensAI
  • AI Inferencing
  • AI
  • Consumer
Share:

Attachments

  • Original document
  • Permalink

Disclaimer

Lattice Semiconductor Corporation published this content on 31 July 2019 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 01 August 2019 01:04:09 UTC