Intel Corp.’s decision to wait until 2011 to support USB 3.0 in PC chipsets will put the wide adoption of the interconnect on hold for a year, said a senior technology manager at a top tier PC maker.
The issue is the second to dog a major USB initiative, following the virtual collapse of UWB-based wireless USB which is effectively dead, said the source who asked not to be named. In its place, interest is now building for 60GHz technology, but separate industry groups need to unite to ensure the future of it, he added.
Without chipset support from Intel for USB 3.0 aka SuperSpeed USB, adoption in 2010 will be limited to “a few high-end graphics workstations and consumer systems,” said the source. That’s because system makers will be forced to buy discrete host controllers for their motherboards, a relatively high cost.
“It’s hard to commit to an emerging technology like this when the key silicon enablers are not making it a priority,” said the source, referring to Intel. “You get into a chicken-and-egg situation,” he added.
The 5GHz USB 3.0 spec got plenty of attention at the Intel Developer Forum last month with a dozen chip, system and software vendors showing products with throughput up to 250MBps.
At the time one source said Intel originally planned to sample chipsets supporting USB 3.0 in early 2010, then shifted its plans out a year. The PC technology manager confirmed that report. An Intel spokesperson said he had not heard of any delay, but declined further comment.
USB 3.0 “won’t get real traction until it gets integrated in the chipsets,” said the PC manager.
That poses a problem for a handful of chip makers rolling out products such as storage controllers for the technology. But it would not be the first time Intel and Microsoft initiative managers have rallied the industry to support a new spec only to have their own key product teams move slowly to adopt it.
The Microsoft and Intel “tech and strategy groups are not always aligned with the product development teams that are in the mode of trying to make revenue and prioritize what to integrate,” the PC manager said.
Intel’s chipset teams are currently focused on supporting Nehalem, Intel’s first processor to use an integrated memory controller. They also are working through a transition to the 5GHz PCIe 2.0 spec.
“They need to prioritize their time and resources on a whole host of things and have to consider the compelling needs for USB 3.0 now versus 18 months later,” the source said.
Lost UWB
Meanwhile the push for wireless USB has “lost its window of opportunity,” said the PC manager, pointing to the closure of many startups and an industry group backing it. Indeed, one market watcher predicted UWB in general will virtually die off by 2013.
“Now with 60GHz technology getting a lot of executive ear time, we don’t believe UWB will gain traction,” the PC manager said.
However, 60GHz is no slam dunk as the next big wireless interface for systems, he added. Contention over the market direction for the technology between the Wireless Gigabit Alliance and the Wireless HD could slow or even derail adoption, he said.
“It’s a discontinuity in the industry, and we are not interested in supporting multiple organizations for one technology,” he said. “The companies in both groups need to take a mature, adult approach and merge the two,” he added.
On the technical front, a handful of 60GHz startups should leverage existing UWB silicon technologies so they can concentrate their efforts on the challenge of designing 60GHz radios in CMOS, he said. Existing 60GHz startups are wasting time and resources designing baseband and media access controllers rather than licensing available IP.
“I’ve seen this movie ten times before,” he added.
Besides SiBeam, one of the early pioneers in 60GHz, Beam Networks in Tel Aviv and a startup called Nitero in Australia are among those developing 60GHz chips.
PC makers believe 60 GHz offers uses for TV, PC and handheld systems in the home and for office PCs that don’t need a wired link to external monitors.
The Wi-Fi Alliance could act as a certification and testing agency for the technology, the source said. He believes if all goes well it could make it into mainstream products in late 2011.
