Monday, February 25, 2008

Chapter and VersaPHY on 1394

Thanks to Richard Mourn from Quantum Parametrics for providing some color about the recently announced VersaPHY specification from the 1394 Trade Association.

Mourn says the spec aims to help a single 1394 link typically used for high bandwidth streaming media apps branch out to cover less demanding apps that might range from speakers and security cameras to sensors.

"If a car seat has an LCD display in the head rest and the display is 1394, through VersaPHY the PHY in the LCD display [also] could service the seat belt sensor, the occupied sensor, the temperature sensor, etc. All of this would be done using the one 1394 cable into the seat," he writes.

The technique works by creating addressing labels so devices can plug into a link automatically. It requires no new hardware, however users need to agree to some standards. An effort is underway to set such standards for power management uses in cars, Mourn says.

10G gets new options

Neterion rolls its X3100 series chips today at VMworld Europe, the first chips to support the NetQueue technology of VMware and the PCI SIG's I/O Virtualization. One analyst said the move could accelerate the slow ramp for 10G Ethernet.

But as reported today, competitors such as ServerEngines and NetXen are poised to leap to the IOV standard and Gen2 Express soon. And they have an edge in supporting storage capabilities such as iSCSI.

Beyond the battle of Ethernet and Infiniband controllers, the competition to define the optimal cabling for data centers and central offices is heating up. Startup Lightwire's release at the Optical Fibre Conference of a new CMOS optical transceiver with low power characteristics is one of many new active cable components to roll out in the past year.

The sheer variety of chip and cable options suggests there are still quite a lot of unknowns about how to best build and wire big back end facilities for 10G.

Wednesday, February 20, 2008

Rx for Moto

Yowza, Eric Broockman's from-the-hip analysis of Moto's cellular problems stings true.

"Ed Zander’s mistake was to believe that a simple reorg or two together with a few interviews and snappy slogans would be sufficient to fundamentally remake a non-competitive corporate culture," he writes in his blog.

He notes how Moto designers slavish copied the successful Razr rather than dare to innovate with follow on products, and recalls how Moto flubbed a chance to co-develop a music phone with Apple before Steve Jobs & Co. went on to define the iPhone.

"The Schaumburg guys and corporate types from the cornfields outside of Chicago knew better than Steve Jobs did. They didn’t get the vision. So, Apple went elsewhere. Ouch. What a lost opportunity," he quips.

There's no quick solution to the deep-seated culture problems at Moto that extended to the non-competitive wireless chip unit at now spun out Freescale, he claims. OK, here's my vote for Eric as the next CEO of the new Moto.

Tuesday, February 19, 2008

HDMI in your hand

The recent Mobile World Congress was ripe with demos of high def video captured on a cellphone and sent to a TV, according to my boss Junko Yoshida, just back from the show. Junko reports on a 65nm HDMI core from MIPS Technologies to make such a link.

The MIPS core competes with an HDMI variant that Silicon Image rolled out at CES. Silicon Image used clever design rather than aggressive process technology to solve the problem.

Junko reports that NXP Semi has licensed the MIPS core. I have noticed other recent cellular chip sets using a full HDMI core. Yet to appear are any design wins for the new Silicon Image approach.

I wrote recently that USB 3.0 might replace HDMI, mbile and otherwise, but other bloggers disagreed. In the march to high def everything, the protected cellular link is perhaps the most bleeding edge of applications.

Thursday, February 14, 2008

Sun still shines on IB

I was humbled when Valley vet Andy Bechtolsheim sought me out after a Sun Microsystems press conference yesterday. I had just asked Sun's systems VP John Fowler a question about unified data center networking and he pledged to ship such products on Infiniband in 2008.

Bechtolsheim wanted to make sure I knew all the technical advantages of IB as a unified net (high bandwidth, lower latency, built in resilience) and the downside of the FCoE effort (standards not done being the biggie).

It was interesting to hear Bechtolsheim point out weakness in his former employer (Cisco) than could trip it up in the push toward unified Ethernet nets, especially given he had made more than a few million when the networking giant bought his Ethernet startup in the boom days.

Just another example that technologies can look very different depending on where you sit at the moment. I expect Ethernet will become the unified net and that it will be slooooow getting there. I also expect Sun will be finding interesting technology arbitrage positions as long as savvy CEOs like Jonathan Schwartz can keep smart folks like Bechtolsheim on board.

Wednesday, February 13, 2008

Steps to converged Ethernet

The vision of a converged Ethernet net in the data center continues to be a moving target, but chip makers are moving toward it. NetXen, ServerEngines and others are moving to 5 GHZ PCI Express and the Express I/O virtualization standard. Their efforts will help deliver better throughput for their 10G Ethernet chips.

But still out on the horizon is the Fibre Channel over Ethernet standard which reported progress yesterday. By this time next year I expect these companies and their counterparts will be announcing support for FCoE.

Meanwhile the software development efforts are “massive,” according to Kim Brown of ServerEngines. Case in point: the company’s chip was complete in March 2006 and its initial software for TOE and iSCSI shipped last month. Oi vay!

Monday, February 11, 2008

Bluetooth hops on the Wi-Fi bus

The bus to the ultrawidband nirvana of Gbit/s+ data rates is taking too long. So the Bluetooth SIG has opted to use the transport independent version of it protocol ride 802.11 as well.

That version of Bluetooth won’t even be available to members until mid-2009, which begs the question of why the group announced this now. My guess is they are trying to hold off interest in other technologies such as 60 GHz radios which had quite a coming out at CES for wired systems. Academics are already working hot and heavy on versions of 60 GHz technology suitable for mobile devices.

Another interesting wrinkle is that the group is only looking to support .11abg and .11n support is not even on the road map for now. That seems to conflict with the stated goal of getting quickly to higher bandwidth, but perhaps broad adoption is a bigger driver than the highest possible speed.

The news amounts to one more sock in the gut for UWB. The wireless USB version has been dogged by low data rates. Regulatory issues mean OEMs need to have different configurations for different markets. Meanwhile design wins have been more a trickle of experiments than the predicted flood.

Diverging prescriptions

At last week’s ISSCC, I heard engineers in medical electronics pitch for just about every form of wireless networking imaginable.

Medtronic said it’s happy with the 400 MHz MICS standard for its implants. Toumaz Technology said it needs a custom design to hit its ultra low cost and power targets. A UC researcher said he needs high bandwidth to capture brain signals and that is leading him to ultrawideband. Another academic said he needs to resolve underlying issues of capturing individual signals before he can even consider wireless.

This diversity does not bode well for an emerging IEEE effort to set standards for body area networks that hopes to embrace not only medical but consumer uses as well. The good news is there is plenty of room for technology development and adoption where medical meets wireless.

Sunday, February 10, 2008

Got a hottie for Hoti?

Believe it or not, it’s time to get in a submission for the annual IEEE Hot Interconnects conference at Stanford which runs August 6-8 this year. The committee is looking for “state-of-the-art hardware and software architectures and implementations” in a broad range of areas including but not limited to on-chip, chip-to-chip and network interconnects for silicon, copper and optical transports.

If you got a candidate send it by March 1 to the link found at the conference Web site. I’d be pleased as punch if you sent an FYI on what you are up to my way as well at rbmerrit@cmp.com.

Thursday, February 07, 2008

Academics crank on 60 GHz

Clearly 60 GHz design has hit the radar screen. Three leading research organizations came to ISSCC talking about 60 GHz advances.

The most significant move came from Belgium’s IMEC which described a digital CMOS receiver antenna to limit high-path signal loss. The dual-antenna device uses a programmable phase shift and integrates a low-noise amp and down converter. IMEC’s next step is to use 45nm technology to craft a four-antenna device that integrates a phase-lock loop and ADC.

IMEC is inviting the industry to join its research efforts. Back in the U.S., Berkeley’s labs have already spawned what appears to be the early leader in the field, SiBeam. But the startup’s initial products are aimed at wired-only systems.

Graduate students from Berkeley and UCLA came to ISSCC showing low-power 60 GHz receivers dissipating as little as 24 mW and thus suitable for battery-driven devices. But the 60 GHz transmitters eat perhaps twice as much power, and both researchers are still in early stages of designing full transceiver to bring 60 GHz to mobile systems.

Pinching pennies

It’s an old trick in the semiconductor game, shrinking costs with silicon integration to compete taking products to mass market. Kudos to Atheros, SMSC and STMicro who gave examples in the last 24 hours of doing this with 802.11n, Bluetooth, USB 2.0 and WiMax.

Atheros presented papers yesterday at ISSCC on what it calls it’s the first 802.11n chip with baseband and radio on chip. It knows competit0ors are not far away with its part which should ship soon. Similarly it described its first Bluetooth 2.1 chip, and integrated part slightly smaller than the one from competitors such as Broadcom.

SMSC today shrinks its USB 2.0 transceiver to new size and power lows, hitting a sub-$1 cost for the first time. The company is talking just a little about nits views on the still-emerging USB 3.0, a topic I’d like to hear more about.

For its part, STM showed the first steps in research toward blending .11n and WiMax into an integrated chip suitable for a cellphone. The company has work to do yet before it pulls this one off, but it got plenty of attention at ISSCC for its efforts so far.

Tuesday, February 05, 2008

Seeking a body-friendly net

A relatively new IEEE group (802.15.6) has put out a call for proposals for body-area network technology. They seek a “standard optimized for low power devices and operation on, in or around the human [or animal] body to serve a variety of applications including medical, consumer electronics and more.”

Tech requirements include support for a quality of service scheme, extremely low power, and data rates up to 10 Mbits/s. In addition, applications need to consider the so-called Specific Absorption Rate of different body types. They must also respect limits for transmit power in US to less than 1.6 mW and in EU to less than 20 mW.

The group aims to attract to its March meeting in Orlando giant medical electronic and consumer companies to weigh in on what they think will be the big apps for a wireless BAN and the technologies required to serve them. “This is a personal area network with the characteristics of the body in mind,” said Arthur Astrin who chairs the group.

I bumped into Astrin in a medical electronics session at ISSCC this week. Plenty of candidates for this work here i9ncluding developers of wristwatches and disposable band-aids that act as wireless monitors, deep brain implants and much more.

It could be a decade too early, but I think this will be one of the most interesting new wireless areas to watch.

Friday, February 01, 2008

Will USB 3.0 overwhelm HDMI?

That’s what at least one attendee was wondering after the first industry review of the draft USB 3.0 spec. A group of as many as 300 came to the two-day, closed-door event in San Jose last week to hear the latest about the emerging Super Speed USB which aims to hit data rates up to 5 Gbits/second.

“I have attended two or three wireless USB events, and they were not as big as this. Almost everyone I know in the supply chain was there,” said Mark Fu, director of marketing for the connectivity group at SMSC which makes USB transceivers.

“There are a lot of things you can do with 5 Gbits/s such as carry uncompressed video, so I wonder whether this will challenge HDMI,” said Fu.

Interesting thought, given the fact last year Intel bailed on the UDI effort Silicon Image created to get HDMI into PCs and mobile systems. Although HDMI is well entrenched in HDTVs, it is looking less like an industry standard and more like a Silicon Image standard every day. And wouldn’t OEMs prefer one link (USB) that handles everything?

External hard and solid-state drives could use the target 300 Mbytes/s throughput for USB 3.0 as an alternative to external serial ATA, too.

“That’s almost as fast as SATA II,” Fu said.

Any challenge is a ways off. The 3.0 spec is only at a 0.78 version, although Intel hopes to have it complete in June.

“The way they talked about the connectors, EMI issues and other concepts you got the impression they have done a lot of work in the lab and this is not just a paper spec,” said Fu.

Indeed, Intel has some basic USB 3.0 hardware running the lab. There was a lot of back-and-forth in the San Jose meeting about cable issues—such as whether to consolidate USB 3.0 to just one micro-B style connector—the basic PHY, link and protocol aspects of the spec “seemed pretty solid,” said Fu.

I said before USB 3.0 will douse whatever heat Firewire is still generating. Now maybe a broader story about USB uber alles is in the making.

I reached out to the Intel and USB-IF folks for more today, but hey it’s Friday. I didn’t hear back. If anyone else at the big meeting has any comments to lend, go ahead and post them here or drop me a line at rbmerrit@cmp.com

Will cameras play tag?

That’s what IMS Research believes forecasting that the market for cameras with integrated Global Positioning Systems capabilities will grow 200 percent on a compound annual basis over the next five years--coming off a very small sub-million unit base, of course. Makers of digital cameras and GPS chips are both hungry to open up new opportunities. But I think the marketing geniuses have more work ahead to flesh out this concept of geo-tagging pictures and turn it into a compelling use case.

Matia Grossi, author of the IMS report, does well to note that camera makers believe GPS chips are “too power hungry, too expensive and take too long to get a location fix.”

New approaches aim to address those issues, according to the report. The Snapspot software from NXP-spinout Geotate turns on a GPS receiver only for a fraction of a second while the user takes a picture. Air Semiconductor’s Airwave-1 keeps the receiver always on, drawing 1mA average and dynamically balancing accuracy with power based on the application.

Don’t expect many GPS-enabled cameras at the PMA conference in Vegas, but there’s enough activity to suggest they will come along eventually, Grossi concludes.

 
interconnects