FPGA reference designs span security, crypto, machine learning
With these reference designs, customers are equipped with additional resources to accelerate the development of differentiated and innovative products. The iCE40 UltraPlus is a highly energy-efficient computing solution for sensor stitching and repetitive number crunching, suiting it for offloading power-hungry application processors in battery-powered devices. Lattice’s small form factor, low cost and low power FPGAs offer enhanced memory and more DSPs to improve system performance and extend battery life.
Built upon existing IP and paired with the new reference designs, the iCE40 UltraPlus underpins a complete solution for enabling sensor-to-cloud security and acceleration at the edge, for performing functions such as data capturing, aggregation, encryption, processing and transmission. Key applications for the new iCE40 UltraPlus FPGA reference designs include:
· Machine Learning / On-device Artificial Intelligence – Employs trained neural network algorithms for always-on, low power human face detection using a low resolution image sensor
· Graphics Acceleration – Supports always-on graphics such as clocks in mobile and wearable device displays at ultra-low power, while the application processor is dormant
· ECC Security – Encrypts sensor data before transmission to the Cloud or setup an authenticated communications protocol between two systems
· LoRa – Controls a LoRa compatible radio to send processed sensor data miles away
Lattice adds, “With improved DSP performance, flexible I/Os and increased memory for buffering, the iCE40 UltraPlus brings added intelligence to smartphones and IoT Edge products, and security to the Cloud.”
The iCE40 UltraPlus is the latest addition to the iCE40 Ultra family, delivering eight times more memory (1.1 Mbit RAM), twice the digital signal processors (8x DSPs), and improved I/Os over previous generations.
next page; comments by Lattice partner companies…
“IoT Edge devices demand energy efficient, always-on sensor functionality, and Lattice’s low power and high performance FPGAs have demonstrated that they can meet the connectivity and processing needed for these advanced applications,” said Terry Moore, CEO at MCCI. “By using the iCE40 UltraPlus with our LoRa design stack, we can support various anomaly detection techniques for always-on security in battery powered applications.”
“To meet mobile product requirements of always-on sensor functionality, while conserving power consumption, we looked for the smallest, lowest power and lowest cost FPGA device to complement our designs,” said Ulli Mueller, vice president, sales and marketing at Think Silicon. “By integrating the iCE40 UltraPlus FPGA with our NEMA|dc display processor technology, we enable graphic acceleration for always-on displays in wearables and other mobile and embedded display devices.”
“FPGAs are ideal for AI and Machine Learning applications at the Edge due to their parallel processing architecture,” said Guy Lemieux, CEO at VectorBlox. “Our binarized neural network implemented in Lattice’s iCE40 UltraPlus FPGA supports face detection in always-on cameras for smart homes, security, and other mobile devices.”