Gfast map

gfast map copy


In 2016, looked very promising.
Thousands worked at developing and deploying.
It wasn't enough. 
Most carriers are investing
in fiber or 5G instead.


Dark Blue: Firm commitments from incumbent: BT (10M), Belgacom, Australian NBN, Swisscom,  Austria, Bezeq Israel, Chunghwa Taiwan, Telus Canada, Telekom South Africa, SK Korea, (U.S.) AT&T, Century, Frontier, Windstream, Belgium, Omantel

Mid Blue: Smaller carriers in Germany, Norway, Finland, Japan

Light Blue: Incumbent likely:  France, Germany, Italy

Double spectrumDeutsche Telekom and NetCologne have tested it. Adtran thinks they will be able to ship soon, but NetCologne doesn't expect much until next year. Broadcom's Greg Fischer confirms to me, "We entered production with 212 MHz G.Fast devices last quarter and also released production standards compliant iDTA.  We’re working with standards groups to assure same with cDTA." 

212 MHz should deliver as much as a gig up to about 200 meters.

That covers most buildings. Almost all would be covered by a second system is deployed for the upper stories.

Broadcom has a "policy of not publishing data on our devices." That goes back to the days of Henry Nicholas, apparently a fan of ‎Sun Tzu. (Henry, a brilliant engineer, is a fan of many things. See this and other articles in the LA Times.) 

NetCologne's demo ran at 1.6 gig down, 200 meg up, close to the predicted maximum speed of the technology. The DT/Adtran was only specified as "over a gigabit." Lacking more details, I wondered whether there was a problem with chip performance. Questions like that are inevitable when information is limited. 

Vectoring noise cancellation is of course a challenge at the high speeds, so everyone will be waiting for demonstrations with a full binder. Australia and others are interested in remote power, also tbd. Broadcom has done innovative work with local error correction, and I'd like to see that shown as well. 

Most important, traditional industry practice is to have a second source. Broadcom has just been locked out of the crucial German market in a patent dispute. They also lost a preliminary decision at the U.S. Trade Agency. They happen to be right on the substance; Tessera is demanding an unreasonable royalty. But that holds little weight in the absurd patent system.

Stuff happens. A factory can catch fire. A huge customer can demand priority. A development program can fall behind. (35b was promised for 2016. DT now says it won't be ready until h2 2018.) Everyone wants to be protected.

Sckipio is hard at work on 212. It will be reassuring when the two demonstrate interop, as the lead at a very large telco discussed with me.

 Here's the pr from last fall


Broadcom Drives into the Mainstream with New DSL Infrastructure Device Family

Second-generation chipset delivers the most comprehensive solution, with industry-leading density, functionality, and power efficiency
SAN JOSE, Calif., and SINGAPORE , October 12, 2016
Broadcom Limited (NASDAQ: AVGO) today announced the availability of the industry’s most scalable platform for DSL service provider infrastructure. With a 4X reduction in device count and improvement in power dissipation of up to 40%, the Broadcom® BCM65400 family of devices enables the creation of a complete portfolio of systems with simultaneous VDSL and G.vector capability, from reverse-powered DPUs to high-capacity DSLAMs.
Improving on Broadcom’s production-deployed previous architectures, the BCM65400 incorporates eighth-generation FirePath® DSPs, a new vectoring core, and a complete set of 1G-to-10G datapath interfaces.  These new resources allow the family to offer the industry’s lowest power dissipation, scale to the highest line densities, and provide the benefits promised by recent amendments, including support for the new 212MHz profile, with aggregate speeds of up to 2Gbps, and dynamic time allocation (DTA).
Broadcom’s solutions offer the unique ability to fall back to VDSL2/ADSL standards, in addition to, with simultaneous, independent vectoring domains, enabling the operator to update central office equipment independently from the customer premise. Building upon this, the 65400 expands the addressable market for to point-to-point coaxial cable, often installed in multi-dwelling units (MDUs), apartment buildings and other high-density housing.
“With our second generation infrastructure chipset, we set out to deliver a scalable platform that system vendors could use for the full range of system densities,” said Greg Fischer, senior vice president and general manager of Broadband Carrier Access, Broadcom. “Aligning this new configuration with our ongoing commitment to build comprehensive, industry-leading support for the new capabilities of the standard, we believe the BCM65400 offers a uniquely competitive solution for our DSL system vendor partners and the operators they serve.”
“As we move from field trials to an expanded pilot, Openreach requires a flexible silicon platform that covers the full range of line densities in our diverse infrastructure and maximizes rate and reach performance,” said Peter Bell, director network portfolio, Openreach, British Telecom’s local network business. “We are encouraged that Broadcom has developed a scalable solution that is aligned with our roadmap.”
“Gigabit services for consumer households have become the gold standard for connectivity, and service providers have focused investment on deploying PON further into their networks,” said John Kendall, principal analyst, Connected Home at IHS Markit Technology. “Even as providers provision their networks for next generation bandwidth speeds, DSL is still a very important market segment. provides a powerful tool for providers seeking to deploy fiber to the most financially sensible point, in order to maximize return on capital investment.”
The Broadcom BCM65400 family of devices is sampling today and available through distributors worldwide.