SAN JOSE, Calif. – By 2027, mobile devices will sport all-digital radios and Gbit/second cellular modems powered by 16-core apps processors running at 5 GHz.
That’s just one of several predictions Broadcom Corp. discussed Thursday (Sept. 6) at a symposium honoring its co-founder and chief technologist Henry Samueli who won the 2012 Marconi Society Prize for his work in chip designs that led to the cable modem.
“Most of the predictions are just linear extrapolations, and you hope for the best, but the disruptive innovations are the ones that really change the industry,” said Samueli in a phone interview after the symposium. “I tried to be as realistic as I could and not too science fiction-y,” he said.
Among Broadcom's other predictions: Radios will hit terahertz frequencies with full digital sampling up to 6 GHz. Home gateways will hit 100 Gbit/s data rates handling multiple 8k x 4k video streams. Ethernet switches will enable mainstream 400 Gbit's rates and Tbit/s peaks.
“The biggest issue we all worry about is the end of Moore’s Law,” Samueli said. “I think we have a reasonable runway to get below 10 nm and that will carry us another 10-20 years, then someone along the line will need to invent something new and engineers just do that,” he said.
In the meantime, Samueli is not sweating the current capacity crunch at today’s bleeding edge 28 nm process. “We probably won’t ramp [28 nm products] until the end of next year, and TSMC and others claim they will have capacity issues sorted out by the end of the year,” he said.
Samueli spends most of his time these days fostering a culture of collaborative engineering at the company that now employs more than 11,000 people. At the technical level that involves “creating common [SoC] methodologies and design flows to make it easy for people to share technology across the company,” he said.
“We have proprietary databases where all our IP is checked in and logged, standards are adhered to and all the documentation needed to use it is available along with people available to answer questions,” he said. “We view that as a proprietary advantage,” he added.
Henry Samueli explored the 2027 horizon at a Broadcom symposium.
I guess predictions are used as a vehicle to wonder about the future, they don't have any intrinsic value...predicting what would happen in 1-day is not difficult, most of the time nothing would do...the same with weather forecast, the safest one for tomorrow is that the weather will be the same, this is the most accurate prediction...Kris
What is the use of such predictions, except for journalists?
Why is there always someone predicting what will happen in 10 or 20 years? Can anyone tell what will happen in 1 year, or 1 day?
The most favorable prediction is about how much money will be earned. The opposite is interesting too: how much money was not earned because some copied something, because of counterfeiting.
It never is about something social e.g. how many people get a job or families earning more money.
If what the employees earn would be on the positive side of the balance sheet, this would look different and change slowly economics, even improve for people I think.
I am compelled to recall an album "Amused to Death" by one of my favourite artists, Roger Waters... we will all be amusing ourselves with 100Gig home gateways, if it ever materializes in the remaining years of my life!
"Home gateways will hit 100 Gbit/s data rates..." amuses me, if one were to take the extrapolation from the 10Gbps example. The standard was ratified 10 years ago and only now we see servers are shipping in volume (though the switches were available 4 years ago but suffered the wrath of excessive power consumption). Most home users have computers with Gigabit connectivity but data ratesare a couple of Mbit/s or worse. Assuming terabit switching at the core by 2020 in similar volumes at 10Gig switches are today, it would be a stretch to see 100Gig at home gateways, even with exponential scaling in speeds... just my cynical opinion!
I agree. I think "it" will be something that the majority of us will not have seen coming. Coffee mugs or shoes or bicycles with embedded processors, or something that sounds just as preposterous as a smart phone sounded 10 years ago.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.