Scalable fusion processors meld MCU, AI, wireless and security - Embedded.com

Scalable fusion processors meld MCU, AI, wireless and security

New fusion processors target next generation IoT devices requiring integrated efficient AI, wireless connectivity, trusted security and long battery life.

Alif Semiconductor has emerged from stealth with a family of scalable fusion processors integrating MPU, MCU, artificial intelligence (AI) and machine learning (ML) plus cellular connectivity and security in a single device to address the AI-enabled internet of things (IoT) market.

Established in 2019 and with $72 million funding already, Alif Semiconductor this week launched its Ensemble and Crescendo product families to target next generation, always-connected IoT products. The company said these products fill the market need for scalable, genuinely power efficient devices that integrate AI/ML acceleration, multi-layered security, LTE Cat-M1 and NB-IoT connectivity, GNSS positioning, and integrated memory to enable design of products that seamlessly integrate into everyday life, whether the processing is done locally or in the cloud.

The Ensemble family is built on the latest generation embedded processing technology that scales from single Arm Cortex-M55 MCUs to a new class of multi-core devices — fusion processors — that blend up to two Cortex-M55 MCU cores, up to two Cortex-A32 microprocessors (MPU) cores capable of running high-level operating systems, and up to two Arm Ethos-U55 microNPUs for AI ML acceleration.

This family of devices contains an advanced secure enclave that provides multiple layers of security, such as device integrity protection, secure identity, strong root-of-trust, and secure lifecycle management. Together with large on-chip SRAM and non-volatile memory, accelerated graphics, imaging, and class-leading power characteristics, the Ensemble family is targeted at smart home products, appliances, point-of sale, and robotics applications.

Meanwhile, the Crescendo family offers the same functionality as the Ensemble family, and in addition adds LTE Cat-M1 and NB-IoT cellular connectivity, optional iSIM for simplified subscriber management, integrated RF, power amplifiers, and a concurrent GNSS receiver for positioning, thus delivering the key capabilities required for next-generation smart city, connected infrastructure, asset tracking, healthcare devices, and wearables, all in a single chip.


Syed Ali (left), and Reza Kazerounian (Source: Alif Semiconductor)

In an interview with EE Times, Alif Semiconductor co-founders Syed Ali (CEO), and Reza Kazerounian (president), explained, “We started out to essentially focus on next generation IoT applications. We thought: what do we need to build something superior, to enable IoT devices with edge processing? Security was also an important consideration – we wanted to build a device that delivered multi-layered security. And we wanted to create a broad range of products based on platform enabling a multitude of applications.”

In summary, Ali said, “We founded Alif Semiconductor because we wanted to provide an integrated next generation solution for developers that need efficient AI, wireless capabilities, trusted security and long battery life. This just didn’t exist until now. We expect the Ensemble and Crescendo families will have a significant impact on the market for next-generation IoT devices.”

Kazerounian added, “The Ensemble and Crescendo families introduce a scalable, highly integrated architecture that lets developers create secure solutions spanning multiple applications, with different functional and processing requirements on a common fabric. The innovative approach we are taking is a game-changer and will disrupt the way developers create intelligent machines.”


Alif Semiconductor’s scalable fusion processor family integrate many of the functions needed for AI-enabled IoT devices with wireless capabilities, trusted security and long battery life. The Ensemble E7 (left) is a quad-core fusion processor, while the Crescendo C7 adds cellular IoT modem, iSIM and positioning. (Source: Alif Semiconductor).

The founders were keen to emphasize the level of integration and functionality that they have developed in the new fusion processors and microcontrollers, designed for performance and long battery life, using its own power management subsystem which it calls autonomous intelligent power management, or aiPM. With many deployed IoT devices being battery powered and battery life is critically challenged when having to also perform local processing, AI/ML, and wireless communication, Alif Semiconductor’s aiPM technology allows fine-grained control of when resources in the chip are being powered. This proprietary technique produces class-leading low-power operation, enabling intelligent devices to last longer on smaller batteries. As an example, when operating at 3.3V, Alif Semiconductor indicates power consumption of less than 1.0µA in stop mode with real-time clock running and wake sources active; and in run mode, the figure is 18 µA/MHz on the high-efficiency Cortex-M55 running useful code from SRAM.

Its aiPM technology is a blend several independent smart power domains that can decide to shut down autonomously when there is no activity; internal power conditioning, sequencing, and regulation (no external power management IC is required); close connection to the interconnection bus fabric; and software configuration. The result is that it essentially powers on only sections of the chip that are needed, when they are needed, and off when they’re not. This makes even the complex quad-core devices behave like small purpose-built low-power MCUs when they need to, to extend battery life.

In addition to aiPM, the multi-core devices are architected in such a way to dedicate the high-efficiency pair of Cortex-M55 MCU/Ethos-U55 NPU to operate at very low power levels while sensing the surroundings (vibration, sound, image). The high-efficiency Cortex-M55 MCU will then wake up other portions of the device (high-performance Cortex-M55/Ethos-U55, Cortex-A32’s, graphics) in an escalated way as needed to execute the workload based on the immediate use case. aiPM takes care of shutting them off when no longer needed.

Ali said validation of the company’s solution is strong, with a good Tier 1 customer pipeline. The head of the Microsoft Azure edge silicon devices strategy, Jerome Schang, said, “The solution that Alif delivers fills a significant gap in the market. We are always on the lookout for the most efficient technology platforms for our edge experiences, and the Ensemble and Crescendo families are very well aligned with our customers’ needs.”

Arm has also endorsed the new product family. Mohamed Awad, vice president of IoT and embedded at Arm, commented, “The next generation of IoT applications requires more intelligent, secure, AI-capable endpoint devices at scale. Alif’s new product families, based on Arm’s proven technology, will unleash the potential of AI and enable developers to create innovative solutions that fuel the continued growth of the IoT.”

Alif Semiconductor employs 200 people across the U.S., India and Singapore, with headquarters in Pleasanton, CA, and SoC design at Irvine, CA; operations are run out of Singapore, and software is developed in Bangalore, India. Ali said, “It was very important for us to own all our IP for cellular, so we acquired a team in India doing 4G.” That Indian team Alif Semiconductor acquired was Mymo Wireless Technology.

Ali and Kazerounian themselves have a successful track record. Ali co-founded Cavium which was acquired by Marvell for $6 billion in 2017. Prior to this he served as VP of marketing and sales at Malleable Technologies, where he was a founding management team member, leading to the acquisition by PMC Sierra. Ali also served as executive director at Samsung Electronics where he started the flash memory and CPU businesses.

Kazerounian is known for his work in the areas of microcontrollers, embedded processing, and connectivity. He was previously responsible for the microcontroller and connectivity business unit at Atmel Corporation until its acquisition in 2016. Prior to Atmel, he headed Freescale’s automotive, industrial, and multi-market solutions product groups, where his responsibility included the microcontroller, connectivity, MEMS sensors, and analog product divisions. Prior to Freescale, he served key appointments at STMicroelectronics, including leading the smart card security and programmable systems memory divisions.

Investors in Alif Semiconductor are also key backers in the chip industry and include Lip-Bu Tan at Celesta Capital (previously WRVI Capital, but rebranded earlier this year), Kleiner Perkins, Lightspeed and Horizons Ventures.

>> This article was originally published on our sister site, EE Times.


Related Contents:

For more Embedded, subscribe to Embedded’s weekly email newsletter.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.