One of the details we have managed to clear up about the Broadband World Forum respective demonstrations of G.Fast capability, is that Broadcom was definitively playing catch up against Israeli upstart chip designer Sckipio and used the very first copies of a new chip on its stand, before anyone else had seen them.
Jim McKeon, senior director of product marketing at Broadcom made no bones about it when we spokes with him last week – “The chips on the stand had not previously been seen by anybody else,” and they were effectively straight out of the factory.
“We used the latest silicon and these chips had not been released to customers, but they are going out to them during November,” said McKeon.
Before this, Broadcom had been using its first run chips that were part of the same product families, but they were pre-production versions, “Typically these are very different, and the earlier chips gave us a chance to get real world experience to help debug them and look for performance enhancements.”
Sckipio and its partner Lantiq have been led to believe that the earlier chips, performing to a lower spec, roughly 500 Mbps, perhaps without fully robust cross talk cancellation, was what it was up against with its new chip range.
“As far as I know,” boasted McKeon, “these are the only chips that have actually performed at the specified 1 Gbps in the G.Fast standard. Sckipio had indicated confidentially around the show that their chip was performing at 700 Mbps, in a cluster of 16 in a bundle, so it required full noise cancellation. Sckipio was under the impression that both Broadcom and Alcatel, who both had G.Fast demonstrations at the show, had only shown older, non-standard chips on their stands, but this was because only the closest clients were allowed in, and journalists and analysts were prevented from seeing the chips in action.
“We had 2 demos, one was a single line from point to point without any noise cancellation working across a 1 meter line, which showed the maximum performance of 1 Gbps, and we also showed 4 vectored lines over 100 meters that operated at 700 Mbps, which McKeon said was in line with the spec. “G.Fast is supposed to provide a benefit up to between 200 to 250 meters and after that it starts to resemble VDSL2,” he added.
To us that sounds analogous to the Sckipio testbed, no better, and was a big risk given that the silicon was hot off the press, and virtually untried.
It now looks to us that Broadcom continues to play catch up, but has certainly made some headway over the lead that Sckipio has built up over the past two years.
We also asked about deployment scenarios, and McKeon said, “At present cross connect boxes are installed at between 300 meters and 500 meters from every home, but copper is already closer than this in most MDUs, certainly under 200m.
“At Swisscom, the tier 1 partner who was quoted in our press release, it has its boxes installed in a man-hole in the ground within 200 meters of each home, so it is very easy for an operator like Swisscom to install this technology on the same boards, alongside vectored VDSL2, which it has already rolled out to its customers.”
“Implementing vectored VDSL2 depends upon a telcos’s loop lengths, and they fall into two categories, those with short loop lengths who will use VDSL2 throughout their footprint to replace the connection at the DSLAM. Those with long loop lengths may find that fiber is a better investment.
“But there are some telcos, especially where they expect to find themselves competing against cable’s DOCSIS 3.1, and they will use G.Fast from the cabinet to remain competitive. In that instance G.Fast is an extension of a fiber plan to reach the home. There are also many local factors that affect the overall expense – labor is a big part of this, so there is a wide spectrum of use cases from installing G.Fast to a light pole in the neighborhood, to putting in the basement to feed an MDU.
“In the US we see this more as a technology which will terminate fiber in apartment units and run G.Fast up inside the building.,” said KcKeon.
He also added, “China was heavily committed to the standards process but is more focused on fiber to the home, because its operators have a stronger say in the rights of way to a customer than in some other countries.
Faultline is covering this so that we can sense early any dramatic shifts away from cable being the leading broadband technology and see any changes in the dominant chip suppliers to broadband.
Right now this industry is in the phase where chip companies are trying to attract device makers, and there will be a lot of trials during 2015 and volume installations will not begin until 2016.