The Internet of Things is getting a lot of attention from such silicon makers as Intel, Advanced Micro Devices and Cisco Systems.
The vendors are looking to supply the chips that will help fuel a market that Cisco officials have said already has helped businesses reap more than $613 billion in profits this year alone, and could hit $14.4 trillion by 2020.
AMD officials on Sept. 9 unveiled the company’s embedded chip roadmap for 2014, including a 64-bit system-on-a-chip (SoC) codenamed Hierofalcon that will be developed in conjunction with ARM. Chips for embedded systems are a key growth area for AMD, and one of the major trends fueling the need for embedded silicon is the Internet of Things—or what AMD calls Surround Computing—the idea that intelligence is being put into a growing range of connected appliances, cars, manufacturing systems, personal devices, wearable computers and other items, and that all these connected systems and appliances are being connected to the Internet and each other and are generating tremendous amounts of data that needs to be stored and analyzed.
The next day at the Intel Developer Forum, Intel CEO Brian Krzanich announced that the company was developing a new family of chips called Quark, which will be one-fifth the size of Intel’s low-power Atom SoCs and consume one-tenth of the power. The Quark SoCs initially are being aimed at such embedded segments as the industrial Internet of Things and wearable computing devices.
Cisco followed that up two days later with the launch of the nPower X1, a network processor that has more than 4 billion transistors, can offer multi-terabit levels of performance and handle trillions of transactions, and is aimed at the Internet of Things—or what Cisco calls the Internet of Everything.
“The Internet of Everything will require extremely advanced silicon,” Nikhil Jayaram, vice president of engineering for Cisco’s service provider business, said in a post on the company blog announcing the nPower X1 chip. “With over 4 billion transistors, this highly integrated 400G bps throughput single-chip will enable Terabit-class solutions. It has sophisticated programmable control using open APIs and advanced compute operations that makes it ideal for software-defined networks while handling extremely high event rates. It will help simplify network operations and allow new business models while it enables our customers to both support rapid bandwidth growth and transform the Internet.”
According to a report released by Cisco earlier this year, there were 2.3 billion Internet users and 12 billion network connections—fixed and mobile devices, as well as machine-to-machine (M2M) connections—in 2012. Those numbers will grow to 3.6 billion users and more than 19 billion network connections by 2017.
In a slide during their presentation, they announced Hierofalcon and other future embedded chips, AMD officials noted that the Internet of Things and Surround Computing, along with “next-generation embedded systems,” will help fuel the next growth in the computing industry. On the chart, AMD officials noted that among the fastest-growing segments in the embedded chip space are traditional embedded systems and intelligent systems.
Intel, Cisco, AMD Eye the Internet of Things
AMD officials are looking to make a significant mark on the entire embedded space, and will do so with both x86- and ARM-based offerings, part of the vendor’s initiative to give customers and users platform options. For its part, ARM in March unveiled the Cortex-MO, which officials said brings 32-bit computing into the same space and a smaller power envelope than current 8- and 16-bit chips. ARM is targeting the Cortex-MO+at intelligent sensors and smart control systems for a wide range of uses, from home appliances and white goods to medical monitoring, metering, lighting and power.
Charlene Marini, vice president of marketing for ARM’s embedded business, noted during AMD’s announcement that ARM last year sold 4.1 billion chips into the embedded space, and that partnering with AMD will help the company extend its reach into an embedded market that VDC Research has said will grow from $11.6 billion this year to $15.5 billion by 2016.
During his keynote, Intel’s Krzanich said the new Quark chips also will help drive the company’s x86-based architecture into new areas. It also will be synthesizable, which means other companies will be able to build their own IP on top of the SoC. Quark will enable the continued development of intelligent systems, Intel President Renee James said during her IDF keynote.
“All of [the new intelligent systems] will be connected,” James said. “All of it will be able to compute.”
ARM’s Marini said it’s difficult to assess Quark right now because there is so little known about it. However, she questioned whether Intel could make much money from such SoCs, given the relatively low margins when compared to traditional server and PC chips. Marini also said Intel will have to play catch-up to ARM and other vendors.
“It’s a large market, it’s a growing market, and we’ve been looking at it for a number of years,” Marini told eWEEK.
Cisco’s nPower X1 will offer 400G bps throughput, with all packet processing, traffic management and I/O functions integrated on the chip. According to the company, it also enables solutions with eight times the throughput compared with Cisco’s previous best network processor, and with a quarter of the power per bit.
Cisco officials are scheduled to roll out some networking solutions that feature the nPower X1 chip during a Webcast Sept. 24.