Thursday, December 20, 2007

FireWire fizzles

We at EE Times have covered IEEE 1394 aka FireWire for years, but this patently good technology may be on its last legs.

I have a story coming in Monday’s paper (I hope) that quotes senior engineering management at both Sony and Moto saying—unprompted--its time to say bye-bye to this interconnect. Smacks of a kiss of death to me.

Brian O’Rourke of In-Stat noted the technology is good but has always been the second or third choice in any given market. It takes a back seat to USB in PCs, HDMI in TVs and set-tops and MOST in cars. A Rodney Dangerfield of interfaces!

Why? It used to be faster than any alternative. But I guess there wasn’t demand for its 400 Mbit/s data rate circa 1993. Then there was the dollar-per-port royalty from Apple that apparently thought this was a product not a technology. Then the FCC foisted it on unwilling cable-TV companies who put chips in boxes without ports. Oi vay!

Kudos to the 1394 Trade Association and their members for their candor and persistence with efforts like HANA. At CES, the HANA 2.0 software will be shown running premium content from NBC Universal from three cable-TV systems across five media types, and the group is talking up its long-planned 3.2 Gbit/s version.

But HDMI seems to have won the TV and set-top, we don’t need another home network type (especially not an interconnect masquerading as a network) and there are plenty of alternative links with speeds as fast, features as rich, power consumption as low and prices driven to the floor thanks to mass adoption--some of them are even wireless.

Sign of the times: I met an EE from one 1394 company at the MoCA conference last month scouting around for new growth opportunities. Maybe there is just one or two more stories to write about this technology—its demise and burial.

Life in Embargo City

One of the hardest parts of my job is dealing with the reality distortion field. There are big things companies are doing—and mistakes they have made--that they don’t want to talk about publicly. Rather than be frank they with withhold or embargo information, ask you to sign NDAs or try to impress you with a content-free marketing tap dance.

Case in point: I know Silicon Image is doing some stuff in mobile and planning some stuff in networking. But there is some stuff I agreed I would not say (yet) and other stuff they will not say yet (unless you are a partner under NDA), and in the meantime a lot of spin in the form of fancy but vague Powerpoint to keep Wall Street interested in the stock.

Almost every company I deal with plays these games as part of doing business. Consortia do it too: Right now the USB-IF is not being fully forthcoming on their DTV-to-mobile plan-in-progress, and the people at Cable Labs and the CEA won’t say much about their networked OCAP and Digital Cable Ready-Plus plans.

Of course, I have my own part in this ecosystem, trying to get stories before anyone else does. It can be fun and useful, but sometimes it’s also frustrating on days when the information lockdown and marketing spin has me reeling. I guess those are the days when the little man behind the curtain who pretends to be the Great and Powerful Oz is having a good day.

Friday, December 14, 2007

The TV plug plot thickens

Add the FireWire crowd to the mix of people trying to define the plug of choice in the digital living room. The 1394 Trade Association says it is aiming its new 3.2 Gbit/s spec at carrying uncompressed video. That's HDMI territory, baby.

Upping the ante, the trade group said it is working on a spec for 1394 over coax. Sounds like a home network play to me. That likely will compete with whatever Silicon Image has going on behind the curtain called the Personal Entertainment Network.

So, we have digital TVs adopting HDMI as a de facto standard, set-tops with FireWire --thanks to the FCC requirement for a copy-protected interface--and the world of PCs, cameras and cellphones wearing various sizes of USB--plus a new USB variant in the works for carrying compressed video between TVs and devices. I love a good mud fight!

Thursday, December 13, 2007

HDMI and USB fight over TV

This is like one of those living room battles over who gets the remote control.

I am told the USB Implementers Forum is working on a version of USB that aims to link mobile devices to TVs. The spec is supposed to be out sometime in 2008. A USB-IF spokeswoman said the new spec aims to carry compressed high def video and should complement HDMI. I don't have much hard data on this, so I welcome any comments on it posted here or at

The interesting wrinkle here is that while USB has got significant traction in digital cameras and is starting to take hold in cellphones, it ain't nowhere to be seen in TVs. That's the terrain of HDMI, and Silicon Image, HDMI's backer, has been showing the financial community foils about how it sees mobile devices like cameras and cellphones as its next big expansion area.

Of course these wired guys will have to tussle with all the wireless options springing up as well—Wi-Fi and its derivatives, Bluetooth and derivatives, UWB and 60 GHz radios to name the main ones. This cat fight alone could make for an interesting CES next month.

The UWB crew reacts

Jack Shandle over at Wireless Designline has a good rundown of the reactions the ultrawideband crew are having to the report this week on the underwhelming data rates for wireless USB. His discussion of the unresolved issues around detect-and-avoid strategies and multi-frequency support is particularly cogent.

I had a chat myself with Stephan Wood, president of the WiMedia Alliance, yesterday for a story I was working on. I applaud his idea of the industry proactively planning use of unlicensed spectrum so a lot of sub-optimal devices don't swap the airwaves.

But I have to note Wood's notion of possibly reserving 60 GHz for uncompressed video and UWB bands for compressed video sounds suspiciously like a way to carve the pie without leaving much for the Wi-Fi that is the best established and perhaps most proactive of all wireless camps these days with its Gbit effort well on its way.

What's more, like it or not, we live in a capitalist techno-democracy where anyone can field anything and consumers vote with their dollars about what wins. That isn't always the prettiest or shortest route and it doesn't always favor the optimal technology, but that's the way the game is played.

Tuesday, December 11, 2007

Bad grades for wireless USB

OK, wireless USB promoters, you are grounded until you get those data rates up. The official tests results from Octoscope are in. They are better than the 20 Mbits/s the group was reporting initially back in October. But the 50 Mbit/s results of the final scores are still not up to par—go back and hit the books, people. Remember, even though you are a big consortium there are plenty of competitors who want to eat your lunch.

Mike Foley of the Bluetooth SIG says his group is still optimistic it can ride UWB to get to a 100 Mbit/s application layer data rate with BT 3.0. The wireless USB problems according to the BT people stem mainly from the USB protocol itself which Mike called "not friendly for wireless, they are too chatty and that impacts data throughput and power consumption."

BT 3.0 will use its existing radio as a control channel, only turning on a UWB radio when big chucks of data are ready to be sent. And BT 3.0 will tap into 6 GHz frequencies as well, he added.

More PAN-demonium

I hear startup Kleer hopes to make a splash at CES in January with its proprietary alternative to Bluetooth used last year in MP3 players from Thomson. But I have yet to see a solid technical analysis comparing Kleer's technology to Bluetooth. If you have some data on this, please post a comment or email me at

I am also still waiting for word on a Wibree killer in the works from Qualcomm. The company said earlier this year it has something in the labs. But I have yet to see any details leak out on it.

STM sees DTV, DisplayPort at its Genesis

Good timing for STMicro today, offering to snap up Genesis MicroChip for an estimated $336 million. The industry is on a significant ramp in LCD TVs and digital TVs, as well as the DisplayPort interface Genesis helped pioneer that will appear inside many of those devices. The acquisition will help STM grab a bigger share of the growing market.

Long term the duo will have some nice integration plays, and STM could use Genesis' DisplayPort technology to expand its business in PCs. In the short term, STM may see an opportunity to reduce some redundant costs and make the Genesis parts even more profitable. But even without cuts, this is a good deal. Genesis reported sales of $191 million in the past year, has top-drawer customers such as Dell and Samsung and a nice pile of some 210 patents.

Friday, December 07, 2007

Express gets a tune up

I got a flame today saying the comment in my previous post about Intel's Geneseo being a candidate for PCI Express version 3.0 was off the mark.

"Geneseo does not exist other than as an Intel marketing term, a term never recognized by the PCI SIG. All of the SIG proposals have seen extensive discussion and modifications" from multiple companies, he said.

The comment led to more back and forth and discussions with another source. It all helped me get a somewhat clearer picture of the work on PCI extensions over at the PCI SIG which I posted as a story on the EE Times site tonight.

Clearly, I have yet to get a sense of many of the details of the new extensions in the works for Express. No doubt I'll be hearing more from the SIG in the new year—if not sooner. Stay tuned.

Wednesday, December 05, 2007

Inside the x86

A week or so ago I asked readers if they had any skinny on Intel's thinking about SoC interconnects. I didn't hear anything back, but the reality is I have a fair amount of background on the subject…and I learned some interesting new wrinkles doing a report to appear at EE Times on Monday.

What I already knew: Intel has announced its Quick Path Interconnect as a upgrade for its front-side bus to appear in Nehalem CPUs starting in 2008. It is only for the hardy few co-processor types who need a fully coherent link. The Geneseo technology it is developing as a candidate for PCI Express 3.0 is the link for everybody else.

What I have learned: In addition to these external on-chip interconnects, there is one or more internal Intel interconnects the company plans to support.

What's more I also talked to Chuck Moore of AMD about their use of coherent and non-coherent HyperTransport links as their main SoC boulevards. Intel and AMD have clearly been courting third parties to hop on their different buses. Chuck said these two pairs of links are probably all anyone will ever see in the emerging world of x86 interconnects.

But "an interesting question is how these two pairs of standards will start to mingle," he said, suggesting multi-protocol links serving multiple purposes.

In addition, "there could be interesting protocol extensions beyond coherency for the next generation of HyperTransport," Moore added. Rather than communicating and synchronizing through memory, devices could link up in ways that "are more optimal," Moore said.

Just remember, dear readers, for this strategic peek into the future of X86 SoCs I charge you all of zero dollars. So, next time I ask for a tip, please drop a line or post a note to keep this little economy flowing, mon freres.

Tuesday, December 04, 2007

Cellular cracks open

The biggest interconnect story of the next few years may be the move to open cellular networks, creating the rough equivalent of a mobile Internet. Jake MacLeod, CTO at Bechtel Comms and an old hand in building cellular nets, thinks it is inevitable today's nets tightly controller by carriers will ultimately become open.

"What Google and other new entrants want is access to mobile customers," he told me in an interview for a story posted today. "They have a right to provide them mobility services that don’t have to be tied to an AT&T, Verizon or T-Mobile."

Google has "a very creative play that will be met with a lot of resistance, but if they are successful they could change the face of communications," he said.

Verizon Wireless took a step in this direction stating last week it would open its network to any complaint device starting in 2008. Whoever gets the 700 MHz spectrum up for auction next week could push the trend further, given the FCC's mandate for open access on at least some of those airwaves.

Have any skinny on the details for compliance with Verizon? Got an opinion or rant on cellular nets? Sound off with a posting here or at

Sunday, December 02, 2007

DisplayPort gathers steam

DisplayPort backers gather in San Francisco this week for the third plugfest for the standard written by the Video Electronics Standards Association. The interface has been slowly gathering steam since a critical mass of vendors officially got behind the spec at CES in January. Bob Meyers, the HP display guru chairing the standard effort, said this year's CES will show an even broader group of proponents, products and prototypes—as well as an update on plans for a two-stage upgrade path.

Foreshadowing the rise of this new interconnect, Integrated Device Technology announced today it has created a new digital display division focused on chips for DisplayPort standard. The company has five DisplayPort parts in the works, but has yet to start sampling any of them.

An IDT spokesman said DisplayPort is not only being used as an external PC interface but is also replacing LVDS as an internal link inside notebooks and LCD TVs. He estimated it could appear in 300 million PCs and 100 million consumer systems by 2010.

Friday, November 30, 2007

Wi-Fi for sensor nets?

That's the case GainSpan made over lunch to me the other day. Zigbee, the startup claims, lacks links to network management software such as HP's OpenView and IBM's Tivoli. Existing Zigbee nodes from the likes of Ember, Dust, Crossbow and others lack the interoperable gateways prevalent in the 802.11 world. And some apps need the 1-2 Mbits/s GainSpan's 11b/g Wi-Fi chip can crank out when needed—and still deliver the 3-5 year battery life required of a sensor node.

I hope to dig into this a bit further next week. If you are working in sensor nets with Zigbee, Wi-Fi or some other net, I'd like to hear your thoughts. Drop a comment or email me at

Thursday, November 29, 2007

Intuitive Networks

I am very familiar with the automated data center concepts promoted by the likes of IBM and HP under titles like Autonomic Computing and Lights Out Computing. But it wasn't until I got a copy of the in-house technical magazine from Bechtel Communications that I had heard about a similar idea in networking.

Jake MacLeod, the affable CTO over there, has an excellent article in the issue describing the concept in detail. He says an intuitive network includes:

* Base station "hotels" that house many types of transceivers
* Sophisticated multiple antenna subsystem
* Sensor networks to help determine when to reallocate resources
* Lotsa network management software

It could take decades to learn how to build really great intuitive networks, Jake suggests. Along the way, engineers have lots to learn about interconnected networks of networks that handle personal identity securely and handle location data across distributed system. There are also big challenges in distributed quality of service.

Sounds like the folks at Bechtel and a few other dozen companies have their work cut out for them.

Tuesday, November 27, 2007

No new wires?

According to a survey released today by Parks Associates half of the homes that link a computer to a stereo or TV use RCA and S-video cables. Hey whatever happened to the no-new-wires initiative?

Apparently what happened was a slew of mediocre and hard-to-install digital media adapters and networks. What's even worse is that only nine percent of homes using broadband had links between a computer and a stereo and only four percent hand links between a computer and TV. In the ago of the iPod and YouTube, the space between the PC and the entertainment center looks like the Grand Canyon.

Monday, November 26, 2007

Battling bottleneck breakers

Rambus and IBM are squaring off on whether it's better to bust through the memory bottleneck with new signaling or packaging technologies. Both approaches look like contenders in the 2010 timeframe.

Our EE Times report today talks about Rambus' plans for demonstrating at its Tokyo developer conference this week a 32x clock multipler of 500 MHz memory channels. The channels enable 16 Gbits/s transfers and ultimately terabyte/second throughput between a microprocessor and its main memory.

IBM has shown proof points for 3D stacks that could enable a great gob of SRAM to sit on top of a multi-core microprocessor linked by fast direct-metal connections. Analysts said that could hit the market in 2010-class Power CPUs.

Other shoes are yet to fall. Sematech said this spring it planned to detail a market roadmap for 3D stacks before the end of the year. Jedec defines the standards for memory interconnects, so it has a major voice here. The good news is there is a diversity of good work on this nagging problem.

Sunday, November 25, 2007

Intel's SoC mortar

Hello to those of you beginning to stray back into business blogdom in anticipation of Black Monday, the day we saddle up again after the nice long Thanksgiving weekend. Hope the break was good for you. I got in some great music, but had to spend waaaay to much time dealing with car issues—alas!

I am starting off my week asking for help. Intel has got the system-on-chip religion as I noted in my last wrap up from IDF. What I don't know is what interconnects they plan to standardize on for linking silicon IP blocks. Maybe Intel itself hasn't decided.

I know Intel will choose a handful of internally and externally developed interconnects, and I know they are reaching out to third party IP block suppliers they want to work with. Got any details to share about their chosen interconnects and preferred IP partners? Email me at with your best insights before you down another turkey sandwich!

Wednesday, November 21, 2007

Of birds and flowers

OK, it's time to take a gratitude break--and start marinating my turkey. I may be a little challenged doing the former job today because I had to just put my car in the shop and may not be able to get to the San Jose Turkey Trot tomorrow to benefit local charities…OK so I have a car to put in the shop, I already made the donation online and I can go jogging here in Campbell anytime I want, thank you.

BTW, I heard a wonderful interview with author and educator Jonathan Kozol on NPR Sunday night. Among other things he noted it is important to make social justice systemic, not just seasonal. He provides and excellent example of someone walking the talk.

I hope you have a good break, find things to be grateful for and a way to give back to your favorite cause. For anyone looking for an outlet, Intel has started a program where it will donate a buck to Boys and Girls Clubs every time someone plants a virtual sunflower in their online garden.

Happy Thanksgiving!

Dialing in on phoneline

I caught up with Rich Nessin, president of the Home Phoneline Networking Alliance yesterday. The interview helped round out the pictures I have been getting lately from the conferences of the Multimedia over Coax Alliance last week and the HomePlug group last month.

My take away is that everybody still has a real shot at stardom, but MoCA seems to have slightly more momentum. Verizon is deploying MoCA as fast as it can, and Comcast and Cox coming on slowly late next year.

HPNA is not out of it by any means. AT&T adds as many as 10,000 new IPTV subscribers using a week using HPNA, and five or six new telcos will jump on the bandwagon in 2008, Nessin said.

I'll believe that when I see it. What's more the group badly needs a second source of silicon now that Broadcom and Conexant have basically dropped HPNA products and are gearing up MoCA chips.

Powerline has cards to play, too. Motorola said last week it expects to get powerline into set-top boxes as soon as the IEEE 1901 standard is set, perhaps in the spring. Europe is the big driver here.

Set-top and service companies are demanding home nets support 400 Mbit-Gbit data rates over the next several years and get integrated into SoCs to drive low cost. None of the three camps have that today, but MoCA may have a slight edge getting there based on last week's conference where I heard about two chip makers saying they will deliver SoCs in 2009 or later. Stay tuned.

Tuesday, November 20, 2007

Can HDMI become a network?

Thanks to a reader of my EE Times blog for pointing out to me some recent slides from presentations created for financial analysts by Silicon Image, the HDMI chip designer. The slides suggest a possible wrinkle in the story I overheard about Silicon Image being in stealth mode with a new concept in home networking.

There is nothing very explicit in the foils. However, a couple slides talk about a Personal Entertainment Network the company aims to get off the ground in 2008. It describes this so-called PEN as "an architecture for securely moving digital media in a home network, [enabling] any device to display to any display in the personal entertainment domain. The Personal Entertainment domain is unique to each household and includes mobile devices."

Hmmm, sounds to me like Silicon Image may be wrapping some new software, and perhaps some additional security features, around its interface chips in hopes of creating a broader service that could tempt OEMs to standardize on their products across a range of devices. The company may even take the notion directly to end users as an extension of its Simplay effort that was initially aimed at making sure HDMI devices worked through interoperability problems.

Clearly Silicon Image is riding the growth curve of TVs and related devices shifting the high def. And they are very vocal about mobile systems of all sorts from digital cameras to cellphones being their next big growth area. But can they carve out a network play from what is essentially a relatively expensive content protection interface loved by Hollywood that some operators and OEMs would rather design around via ultrawideband or 60 GHz radio links? This could be a bridge too far, especially considering the company has gone through some recent management changes. Stay tuned.

Thursday, November 15, 2007

Heat's on for Moca, too

The folks at the Multimedia over Coax conference were mum on their road map this week except to say that they have kicked off a MoCA 2.0 effort. Their customer base, however, was somewhat more vocal.

"The next-generation MoCA spec needs to support about 400 Mbits/s within two years and a Gbit in four years," said Mark Wegleitner, chief technology officer of Verizon Communications in a keynote speech. Verizon is installing fibre-to-the-home systems that can hit up to 400 Mbits/s today, and it can't let the current 175 Mbit/s MoCA 1.1 technology be a bottleneck for long.

Jed Johnson (right), senior director of systems engineering at Motorola which supplies set-top boxes to Verizon, echoed the call for more bandwidth. "There needs to be a gigabit path in the home," said Johnson.

What with startup Gigle Semi and the ITU standard in the works potentially delivering a Gbit/s PHY next year, MoCA has some pretty clear marching orders.

Wednesday, November 14, 2007

What's cookin' at MoCA Con

The fall in Austin is sweet and so are all the stories rolling out of the first MoCA conference here. Comcast and Cox suggest they will roll out MoCA networking in a small way starting late next year, and chip makers agree that in the fragmented and confusing world of home networks, that's good news.

But off on the Texas horizon are a few new wrinkles. It seems everyone would like to see a single home network for all copper lines. Startup Gigle Semiconductor (Edinburgh, Scotland) hopes to offer just that sometime in 2008. Founded by execs from DS2 and STM, the company claims it will offer competitive, integrated silicon with two channels—one for powerline and the other handling any copper link (telephone lines, powerline, coax or a combo of any two) at PHY rates up to a Gbit/s. We shall see.

Over in Europe the ITU has been trying to hammer out a standard for a home net that could run on any copper media. Called, it has attracted participation from a broad group including Alcatel, Broadcom, BT, Gigle, Intel, Intellon, Panasonic, TI and Siemens. I'm looking for more details on this effort (what is the target data rate, QoS and status of the spec) if anyone can chime in with info.

One last hot tidbit: Silicon Image, the company known for its HDMI silicon, has kicked off some sort of home networking initiative of its own. I don't have any details about it but I'm hungry to learn more. Post a comment or drop me a note at

Gearing up silicon networks

Kudos to IEEE Micro for a solid issue on the topic of on-chip networks. The best article for my money is the report on the December 2006 Stanford workshop on this topic.

According to the report, the gathering identified a laundry list of technical challenges. Chip networks require at least ten times too much power and have far too high a latency. For example, they scoped out a theoretical chip design in the year 2015 using 256 cores and concluded it would use 20 percent of its 150W budget just on linking the cores.

What I thought was most useful here was the research agenda the group scoped out based on its analysis. They called for new encapsulation methods and libraries in design tools because the circuits and architectures used for silicon network are not compatible with today's CAD flows. Designers also will need formal verification methods and queuing analysis tools to replace simulation which will become inadequate.

They also said we need work on new low voltage signaling technologies, 3D stacking to reduce the length of on-chip wires, on-chip photonics that can be 15-20 times faster than today's wiring and tools that more accurately model traffic on these silicon systems. It's an ambitious agenda, but without these advances engineers won't be able to create tomorrow's multicore processors.

Monday, November 12, 2007

Déjà vu for Advanced Switching

Well, not exactly. Intel washed its hands of an effort to tailor PCI Express for the comms market, and they don't seem to be revisiting that decision.

However, as part of a deal selling off its IXP network processor technology to Netronome, Intel did make a point of insisting the startup adopt Intel's upcoming Quick Path Interconnect (QPI) for its next-generation IXP-based processors. QPI is Intel's answer to AMD's HyperTransport, a 6.4 Gbytes/s link replacing the Intel front-side bus in CPUs starting with the Nehalem generation in late 2008.

Don't expect QPI to try to replace tried-and-true network processor interfaces such as the SPI 4.2. It probably won't even give Interlaken, Cisco's proprietary interconnect, much of a run for its money. Nevertheless, Intel will leverage the link to try to get more design wins for x86 and IXP parts in comms.

Data centers call the cable guy

Now that data centers are gearing up for 10 Gbit and faster systems, people are starting to wonder how to physically link these boxes. Standard optics are too expensive, Infiniband CX4 cables are too limited in distance (and have other issues, too) and 10GBase-T links are too power hungry.

Enter a new assortment of active optical cables including Laserwire, a 10 Gbit serial option announced today (Nov. 12) by Finisar. At as little as 500mW per port, Laserwire is much lower power than even the most optimistic 10GBase-T projections for 2008. It is rated at 10 Gbits across 35 meters and may stretch further in the future. And it doesn't have the problem of no defined standard for powering a link like the Ethernet over CX4 option.

Finisar is throwing its hat in the ring even though it is not ready to say exactly when its product ships or at what price. That's because Intel and startups Luxtera and Quellan already have rolled out three other options to solve the 10G+ cable problem. Each approach has its trade offs.

The fact that they are all jumping into the market indicates there is a clear problem with no single clear solution yet. Watch this space.

Got a beef with your cable guy? Leave a comment here or drop me a line at

HDMI goes mobile

Silicon Image, Inc. has lowered power consumption on its latest HDMI transmitters, the SiI9022 and the SiI9024, designed for mobile devices such as digital cameras, portable media players, camcorders and mobile phones. The 85 MHz and 165 MHz parts can support resolution up to 720p and 1080i, consuming 50 mW for 720p/1080i and 80 mW for 1080p. The transmitters also reduce cost by integrating some controller functions. The chips are priced at $4.15-$4.84.

Back at CES in January, top computer makers such as Dell, HP and Intel threw their weight behind DisplayPort as the next-gen digital display interconnect with content protection. That left the world of TVs and other CE devices to HDMI.

With this move, the HDMI camp could extend its reach to tens if not ultimately hundreds of millions of new devices, but I don't expect it to help them make in-roads into notebooks and desktops. By the way, It's been nearly a year…so where are those DisplayPort chips and systems using them?

Sunday, November 11, 2007

Tracking 10G Ethernet

There will be a lot of hubbub about high-end interconnects at the Supercomputing 2007 conference this week. So I asked how many systems on the Top 500 list use 10 Gbit Ethernet links. It turns out, no one knows yet because the folks that compile the list still collect the data by asking whether interconnects are at the 100 Mbit or Gbit level, a sore spot they are moving to heal.

Meanwhile market researchers such as Dell Oro Group and The Linley Group estimate of the eight million servers that shipped in 2006, 90 percent used Gbit Ethernet. They believe about one million 10G ports or 50,000 10G Ethernet cards will ship in 2007, mostly in high-end Unix systems. That ain't much by Ethernet standards. Nevertheless, the same market watchers say Infiniband is growing about 40 percent a year and should hit sales of a million units in 2010, a breakneck pace for the more niche-y interconnect.

Friday, November 09, 2007

The Friday Crock Pot Report

My apologies to readers of this blog for practically abandoning you all week. It has not been for a lack of things to say, but most of them have been embargoed for Monday's Supercomputing 2007 conference in Reno.

Much as I would enjoy the drive through Tahoe and to the Nevada desert, I will be in Austin next week at the MoCA conference. So look for some home net news here next week as I learn what I can from Comcast, Intel, Cisco, Verizon and other presenters there.

Meanwhile here's what's in the slow cooker this weekend getting ready to be served up on Monday morning:

News about Infiniband, 10 Gbit Ethernet and new options for trying to link up the 10G stuff emerging in the data center. I'll just note that all the emerging active cable options are an indication of a problem I don't think anyone has quite solved yet.

There will also bit little tidbits about HDMI improvements—hey, where are the DisplayPort chips?—as well as news on Intel's QuickPath Interconnect in comms and the latest revision of the Top 500 list.

I never even got to this week's news that included a new 10GE switch from Broadcom and Express Gen2 switches from IDT. And kudos to Atheros, one of the few Wi-Fi startups to cross the chasm. This week the company rolled out a single-chip 11n device and an integrated Bluetooth chip as well as a Gbit Ethernet controller. You've come a long way, baby.

So, here's hoping I get time for a little fun this weekend in between catching up on the interconnect news.

Monday, November 05, 2007

Give me iWarp speed, Scotty

Infiniband has a pretty comfortable position as the growing king of cluster interconnects right now, according to Bob Wheeler of market watcher The Linley Group. Gbit Ethernet has peaked in the Top 500 list of supercomputers, IB is stealing market share from Myricom and Quadrics and the 10G Ethernet bandwagon is bogged down in the mud, according to Wheeler.

He notes that neither Dell, HP nor IBM have started using iWarp chips from the likes of Chelsio and Net Effect in their mainstream servers yet. "Until the big OEMs start investing in the iWarp ecosystem, it is not going anywhere," Wheeler said.

I think I wrote my first story detailing the folks pulling together the iWarp standard back in 2001 or so. HP and other major OEMs were among the big backers then, but today HP is collaborating with Netxen, a 10GE vendor that won't support iWarp until next generation products come out.

Making matters worse for the 10GE crowd, Myricom is rolling out products using its own protocol over Ethernet, and Mellanox plans to support Infiniband on top of Ethernet as an alternative to TCP. "They are fragmenting an already small market," Wheeler said.

The flip side of the coin is that IB is not likely to extend far beyond high-end clustering apps, Ethernet will dominate LAN connections and 10GE will eventually go mainstream once 10GBase-T prices come down in a year or two. Chip designer Fulcrum is giving 10GE a little nudge forward with some new products rolled out today.

"Based on the numbers I've seen from systems integrators, it seems like IB has a significant price/performance advantage over 10GE now, but that will change rapidly as 10G E ramps into volume," Wheeler says.

So, at the end of the day, the IB folks may be crying "Beam me up, Scotty!"

Fibre Channel spinning down

I was briefed today by a storage array vendor switching from Fibre Channel to Serial Attached SCSI drives because, as they see it, that's the direction of the future in high performance, high reliability hard disks. More details next Monday when the embargo handcuffs come off, but it's worth noting the array vendor will keep using FC as an external link. When I checked in with a Seagate spokesman he said they see the same trend.

"FC will be around for years into the future because of the investment that companies have made, but it represents a fairly small portion of the overall interface choices...Although there will be a few companies that remain staunch fans and the FCIA pushes for higher speeds, SAS will grow to be the primary mission critical interface," he said.

Thursday, November 01, 2007

The year of mobile maps?

For years, people have been predicting a world of location-based services thanks to a proliferation of GPS embedded in mobile systems. Maybe it will finally happen in 2008.

One of my colleagues over at the Automotive DesignLine notes that GPS is on a rapid rise in cars. And today iSuppli Corp. claims we are just past the start of the hockey stick ramp for GPS in cellphones.

Well, maybe. The FCC took a go-slow approach to its E911 mandate once seen as a driver for GPS in cellphones, and most handset makers used triangulation to save the costs of a full GPS implementation. But Qualcomm has been pushing hard on GPS since 2000 and Sirf Technology convened a summit on location services recently.

So maybe this is about to happen. Or maybe not. I know friends with their new iPhones have been particularly pleased about how that handset helps them access Mapquest on the road…so maybe handsets with real Internet access will supplant GPS phones before this trend really takes off. I vote for the latter. Why recreate something that's already working?

Tuesday, October 30, 2007

Beyond .11n

Last May the 802.11 group convened the Very High Throughput Study Group to scope out what comes after .11n. One motivation for starting the group was to respond to a call for proposals for Gbit-class wireless networks from the ITU. Craig Mathias of Farpoint Group, one of my favorite wireless analysts, believes the group will ultimately opt for some form of 60 GHz networking, but we shall see.

For the last few months the group has been listening to presentations from AT&T, Intel, Motorola, Nokia and others. I haven't looked through them all, but as part of a story that mentioned VHT SG I did read a couple of them.

The one that struck me the most was from an AT&T Labs researcher, calling on the group to get involved in Terahertz-class regulatory decisions underway in Europe that could cut commercial interests out of that spectrum band. Terahertz transceivers are already in the works from startups such as Phiar Corp. the presenter noted. Like my colleague Patrick Mannion told me years ago—before we even started talking about .11n—we are going to be writing a lot about this wireless stuff for years.

Saturday, October 27, 2007

More rumblings on powerline

By the way, the folks from powerline specialist DS2 did call me last week, and it turns out things may be a little less clear than I thought. Private discussions are still underway on all three aspects of the pending IEEE 1901 standard--access, in-home networking and co-existence, they say. As I reported this week, DS2 also is revving up a 400 Mbit/second physical layer technology that they will demonstrate in New York in November and have available for 2009-class products.

DS2 is putting on a big last-minute push in the face of strong votes that came down earlier this month for joint HomePlug/Panasonic proposals in access and in-home networking. Group chairman Jean-Phillipe Faure said he would not be surprised if a final decision is set by the March meeting.

Anything could still happen at this point, so this is a story well worth watching for the next few months. But it does appear the HomePlug folks are now in the cat bird's seat.

Thursday, October 25, 2007

UWB in the crosshairs

Kudos to Jack Shandle, editor of the Wireless Designline, for encouraging ultrawideband chip makers to get tested. Since Computex in June the word on the street has been that UWB products are getting a fraction of their promised throughput, and now its time to find out the truth.

Shandle has been rallying support for a test program sponsored by UWB competitor Pulse~Link and carried out by Octoscope, an independent lab. So far, Octoscope discovered a Belkin F5U301 wireless USB hub using Wisair and WiQuest chips gets less than 20 Mbits/s throughput at a range of up to 20 feet. A model F5U302 got up to 60 Mbits/s at two feet, the fastest of the products tested. An IOGear hub got 30 Mbits/s maximum and lost its connection at 13 feet. By contrast, transfers using wired USB links achieved 150 Mbits/s in the tests.

Vendors and at least one consultant say it's just the teething problems of early chips and drivers. But tester Fanny Mlinarsky says there must be something more afoot and she has plans for more thorough tests to get to the bottom of the issue.

Ironically, Intel Corp. announced earlier this year ambitious plans to push the wireless USB spec to a theoretical throughput beyond 4 Gbits/second in a version 3.0 now in an early draft. The UWB folks better get to 200 Mbits of MAC throughput before they burn much energy talking up multi-Gbit plans.

The folks in 802.11 land are preparing to eat UWB's lunch with their Very High Throughput Study Group investigating what's beyond .11n. What's more the 60 GHz crowd gained big backing from Big Blue this week.

UWB beware: There will be plenty of wireless local- and personal-area nets on the scene in the next few years, and the competition will leave little room for approaches that can't deliver on their promises.

Wednesday, October 24, 2007

HomePlug surges forward

The results are in from Boston last week. No not the World Series, the IEEE P1901 meeting. Geez, guess I am kinda geeky.

Word is the proposals from the HomePlug/Panasonic/HiSilicon crew got the nod. They want to create a mandatory "interPHY protocol" so HomePlug or Panasonic powerline nets can sense and avoid each other. An optional part of the spec is to support both PHYs so the nets can actually interoperate. Interoperable networks, what a concept!

The proposal still has to get a 75% vote when the group gathers again in San Diego in December. (Here's hoping you are all safe from those horrible fires!) But backers say the three-quarters majority vote won't be hard to get.

I'm having trouble getting a full picture here because the DS2/UPA folks aren't talking to me. And no one has clarified what the coexistence proposals that got tabled at the Boston meeting are all about.

In the long run I have to wonder how much all this matters. HomePlug people will keep shipping their products and DS2 people theirs. Will consumer behavior shift in any way because there is a standard? Methinks not.

Hopefully tomorrow I can make time to catch you up on some UWB and Wi-Fi news, but for now it's late, I am tired and there is plenty to do tomorrow.

Tuesday, October 23, 2007

Phoneline calls home

CopperGate Communications is crowing this morning that it has shipped more than two million of its HomePNA 3.1 CopperStream chipsets to OEMs such as Motorola, 2Wire and Scientific Atlanta. The technology owes a debt of gratitude to telcos such as AT&T driving into IPTV and pulling with them the familiar phoneline approach from the DSL days.

Kudos to this crowd who have toughed out plenty of up and down days as home networking slowly emerges. But the toughest part is yet ahead.

The home network is increasingly a heterogeneous environment. Next month CTO-class speakers from Cisco's Linksys group, Comcast, Motorola and Verizon will discuss their plans for Multimedia over Coax at an event in Austin. Earlier this month, a similar group talked about their plans to use powerline technology sponsored by the HomePlug group. And all this gets blended with the many Wi-Fi and alternative wireless products coming into the digital home from retail.

The home network is happening, but it is emerging as a patchwork quilt. That will open up whole new problems and opportunities over the next five years.

Monday, October 22, 2007

Stirring the wireless stew

Today sees the debut of no less than two new wireless personal area networking options and a proprietary spin on Wi-Fi.

In PANs, my colleague R. Colin Johnson relayed a story about how IBM has teamed up with Taiwan's MediaTek to field a 60 GHz chip set for wireless personal area networks. (I am not finding Colin's story online this morning, but see Cnet's version here.) The news hits the same day startup Radiospire emerges from stealth mode with its 1.7 GHz chip set.

IBM will build an RF chip in its silicon germanium process that MediaTek will pair with its baseband chip, both targeting the IEEE 802.15.3c standard. The link is expected to deliver multiple Gbits/second over 5-6 meters. It will compete with the WirelessHD effort launched earlier this year by startup SiBeam which is evangelizing the technology among OEMs with an ad hoc standards group.

Separately, Radiospire Networks announces today its AirHook chip set using the 1.7 GHz band to deliver throughput of 1.6 Gbits/s. The company loosely refers to its effort as wireless HD, but as far as I can tell it has no relation to the SiBeam or IBM technology. It's not clear to me whether it uses ultrawideband or some other approach.

But wait, there's more. Startup Avnera Corp. is announcing its AudioMagic chip set that uses a proprietary spin on 2.4 GHz wireless networking to send audio around the digital home in a way that it claims is superior to the omnipresent Wi-Fi. Hey, I heard that story before from a startup that wilted on the vine trying to get buy in for a unique spin of .11a for TV makers.

According to Colin's story, Avnera already has some market traction. The company claims it has design wins at a dozen major audio component makers, including Panasonic, and that foundry partner Jazz Semiconductor is in volume on the Avnera parts. Companies including Acoustic Research, AudioEngine, Best Buy and SDI Technology already have products out that have been quietly using the Avnera chips.

The startup claims its chips maintain 30 millisecond latency or less and solve error correction problems without needing to use retransmission techniques used in the more data-oriented Wi-Fi. Colin's story suggests some details about how they do it, and notes the downside is it means Avnera's chips won't work with anything else on the home Wi-Fi net.

Amazing, one day brings three new options for wireless in the digital home. This stew is getting pretty thick.

Friday, October 19, 2007

Another view of the flash drive future

Startup Fusion-io was making some interesting claims when it rolled out its first flash memory card for the PCI Express bus. The card hit 100,000 IOPS which Fusion and at least one competitor said was a new high water mark. A spokesman fro Hewlett0-Packard also voiced support for the Fusion-io card in a press release from the startup.

Fusion's CTO David Flynn said the next big move is getting operating systems to work on an overdue overhaul of the old block storage I/O structures to push to a million IOPS. That will open new doors for packing the equivalent of a storage network inside an existing server blade system with the card.

I ran those claims by Michael Krause, an interconnect expert at Hewlett Packard. His initial take was that the startups performance claims could be further detailed, OSes are moving ahead in I/O, but more needs to be done. Here's a short digest of Mike's reply:

"About 8 years ago, I spoke at a conference and showed moderately high-end servers at that time doing 1 million IOPS. Without additional information on the actual performance workload, it is hard to say whether this is really impressive or not.

"Any new technology that mandates software changes, especially in a major OS subsystem, often sees significantly slower adoption than the advocates believe will occur.

"There has been significant advancements in OS I/O structures including major advances in file systems and storage subsystems. All OS support the class driver model which allows storage technology to evolve rather transparently to the OS, [and] nearly all OS also support the ability to stack or substitute file systems without impacting the rest of the OS.

"It's true that SCSI hasn't changed in years, but that has more to do with the customer demand to keep the basic block storage model intact… There is agreement among many within the industry that the traditional block storage model is limited and that perhaps moving to object based storage is the next logical progression."

Thanks, Mike.

Tuesday, October 16, 2007

Round up for RIO

Thanks and a hat tip to Tom Cox, head of the RapidIO trade group, for a dash of news and humor after a looong day.

The news: Tundra Semi has rolled out a new RapidIO switch that bridges a bank of RIO DSPs to PCI-based host processors. It also has a special port to link to FPGAs while we wait for the FPGA crew to develop their own embedded RIO cores.

That's the only news of note I have heard from the ATCA Summit in Silicon Valley this week. I am away on business in Vancouver enjoying this town despite the rain. If I am missing anything else drop me a note at or leave a comment here.

Tom also pointed out that the processor AMCC released at the recent conference was its first with native RIO. The RIO ecosystem, such as it is, will double is size in 2008, he promises.

And the humor: Yesterday, I mistakenly referred to a home networking group as the Multimedia over Cox [sic] Alliance. The ever energetic Tom informed me he has not spawned any new interconnect initiatives but would gladly accept a nomination if anyone submits his name.

Monday, October 15, 2007

FCoE in high gear

The drive to craft Fibre Channel over Ethernet products is in high gear. Witness today's news that longtime card maker Emulex is teaming up with Cisco spin-out Nuova Systems to co-develop silicon.

The interesting bit here is that the pair already has working hardware under development testing in the lab (probably just an FPGA), expects to have qualification chips early next year and actually to be shipping products before the end of 2008. That's pretty darned fast given the T11 spec is not expected to be in draft form until April at the earliest, and the related IEEE 802.1au work may take even longer.

My take away is that everybody sees a real need here and the data center sector is in rare form putting aside competitive issues to make something happen. So who else out there wants to talk silicon for FCoE?

Sunday, October 14, 2007

MoCA gets frothier

I got a tip from a representative of the Multimedia over Coax Alliance who attended the HomePlug conference last week. The big news is the MoCA crew is announcing a version 1.1 spec soon. It will enable throughput of up to 175 Mbits/second at the media-access layer and support parameterized quality of service.

An Echostar rep also at the meeting said they pressed hard for the parameterized QoS, a capability that allows bandwidth guarantees for high priority apps like delivering that HD stream of the Super Bowl. Powerline and wireless technologies can't do that, he said, because they do not control access to their medium.

Also, Broadcom and Conexant are confirmed to be at work on MoCA chips, though no one has seen any sampling silicon yet. Word is MoCA parts may be a bit pricey compared to slower rate home net technologies, but it’s the throughput and QoS service providers demand.

I expect to attend a MoCA event next month, and hopefully there will be more news there. Meanwhile, as I said before, the home networking story will be a clash of service provider technologies and the Wi-Fi and powerline stuff consumers buy at retail and put in themselves.

Friday, October 12, 2007

Another surge on powerline

Just when I thought it was safe to go on to other topics (look for a Fibre Channel over Ethernet update Monday), there's more noise and data surging about powerline.

Chano Gomez, vice president for technology and strategic partnerships at DS2, blasted out an email today that didn't name names, but was clearly referring to the HomePlug/Panasonic proposal up for discussions at the IEEE 1901 Boston meeting Monday.

"It's the worst of all worlds - it's a Frankenstein standard," he wrote.

"Two or three separate PHY and MAC layers and two or three modems in a single chip…will certainly not supply interoperability. At best it will produce coexistence and is immensely short-term. It's a 'fudge' with all the cost implications that this entails," he added.

As I wrote here and at EE Times this week, the HomePlug/Panasonic proposal mandates a protocol that can detect either company's PHY and insure the nets do not interfere, but supporting both PHYs to share data is optional. DS2 and its supporters have also submitted a spec that aims primarily at co-existence. But now that a potential vote—that HomePlug could win—is approaching, the rival group seems to be singing a new tune, calling for one PHY that all sides agree to even if all have to rev their chips.

Oleg Logvinov of the HomePlug group suggests the DS2 proposal is in bad faith because the group declined to participate in the HomePlug AV standard. Of course, DS2 had a big market edge back then with customers shipping its 200 Mbit/s technology in April 2005, about two years ahead of HomePlug which had a coming out party for its 200 Mbit/s technology this week.

The proof is in the pudding, and that means all eyes will be watching for any flying Boston cream pies next week. Hopefully cooler heads will prevail.

One last note on powerline

At the HomePlug conference this week the chief executive of RadioShack gave a no-nonsense talk about what people pushing new technologies like powerline need to do to succeed. One element was addressing and articulating a real user need.

For Oleg Logvinov of Arkados that need is whole home audio. He showed his HomePlug 1.0 chips with a pair of direct-drive speakers from Tatung delivering music remotely from an iPod dock in another room. Wi-Fi has trouble doing this, especially in European homes with thick walls, and the powerline approach is not as dependent on a PC as a controller as Wi-Fi, Logvinov said. But Wi-Fi can't be having too many problems in this space, given Tatung and Roku are delivering products that meet a similar need.

Arkados has tailored its silicon to support this app with appropriate interfaces and an ARM core to handle audio processing. This is a pretty cool product, but I question whether the market is big enough to be the pot of gold at the end of the rainbow for the startup that has toiled on this path seven years. With a $200 bill of materials, the speakers are a bit expensive for the casual iPod-using teen and I wonder if the HomePlug 1.0 speeds are adequate for the golden ears of the audiophile crowd. A Philips engineer on hand said Dolby audio requires latencies of less than 20 milliseconds, a capability it could be hard to insure on powerlines.

Powerline's in-house battle

Powerline proponents gather Monday at an IEEE 1901 meeting in Boston to see if they can hammer out a common spec—and sparks may fly. HomePlug and Panasonic have a proposal that mandates coexistence between them but makes interoperability optional. The DS2 crowd is calling for one PHY that everyone supports, even if it means new silicon. I'm convinced the DS2 approach is the right one, though I am not sure how sincere anyone is about biting this bullet. Stay tuned.

Thursday, October 11, 2007

Plugging into Zigbee

In this week's announcement of a new Command and Control spec under the HomePlug Alliance, I missed the bigger picture that the HomePlug group is taking on Zigbee in a pretty big way. A lot of the technology behind the new spec comes from Israeli startup Yitran which has a new 180nm chip in the works for the technology set to sample next year.

Avner Matmor, Yitran's chief executive, told me over lunch yesterday that he sees the two technologies as complementary, but clearly they have similar applications, costs and data rates. Yitran's 2008 module will have a bill of materials of about $3 initially and handle up to 7 Kbits/s over powerline, aimed at white goods, lights and alarms for a mix of OEMs and utilities. Matmor said he hopes in the future there will be a common protocol for Zigbee and the powerline technology, enabling hybrid wired/wireless home control nets.

In a way, powerline is coming full circle, back to its roots in X-10 home automation. But Matmor notes the difference is the new technology supports full networking capabilities and will handle Internet connections.

There's a protocol spec for the HomePlug C&C spec that won't be complete until the end of the year. Meanwhile, the rival Universal Powerline Association says it will complete a similar spec based on its technology sometime in 2008.

Look out Zigbee fans. The competition is powering up.

Wednesday, October 10, 2007

A patchwork home

Headed down to the HomePlug annual convention today, here are my views about home networking:

The Multimedia Over Coax crowd has a line into the cable TV providers who will likely start specifying their technology in next-gen set-top boxes soon. (Expect to hear more on this from the MoCA convention in November.) The IPTV folks such as AT&T are already specifying Hone Phoneline (HPNA) because it's mature and uses the familiar media and technology they became accustomed to in the early days of DSL routers. The powerline folks are getting some traction in retail because the Linksys and NetGears of the world have been shipping easy to use, low cost products, but those same folk have been spawning an even bigger surge in home Wi-Fi which rules the roost in home networking today and likely will well into the future.

These dynamics are setting us up for a lot of contention between different technologies. The biggest dispute on the home front will be between carrier-approved secure and nets with solid media delivery like MoCA and HPNA and the renegade Wi-Fi. It will take years to see how all this plays out, and it won't be pretty.

Monday, October 08, 2007

Another optical option for IB

Startup XLoom Communications, Inc. (Tel-Aviv) officially joined the crew offering an optical option for Infiniband today. It's AVDAT 4X takes CX4 connections up to 100 meters, competing with alternatives from giant Intel and startup Quellan.

The most interesting thing to me is that IB is projected to ship a million ports in 2008, quite a jump from about half that this year. No wonder multiple companies are trying to ride its coat tails. It's not vast like PCI or Ethernet but it's substantial.

The big flash pile on

Like so many instant trends in computing, everybody these days is trying to be the top of the heap in the big pile on to pack flash into PCs. Two novel approaches emerged just today.

I talked with David Flynn of Fusion-io late Friday who sketched out the company's push to update the guts of server block storage so the company can help "dissolve the SAN back into the server." It seems to me these folks are on to something significant. Look for more on the details of the Fusion controller and reactions to the startup's effort here and in print next week.

Separately, Insyde Software and Silicon Storage Technology tipped their plans to create a platform for flash in notebooks that other companies could build on. But I am skeptical about their FlashMate approach because the value of the new apps they sketch out seems ,marginal to me and the two companies have a lot of heavy lifting to do to create this platform and rally support for it. That's to say nothing about avoiding the ire of the giant in Redmond.

On thing is for sure, there's plenty of energy going into packing more flash into the PC over the next few years. Stay tuned.

Thursday, October 04, 2007

Hard(ware) times at Microsoft

I sat in on a panel discussion at the ARM Developers Conference yesterday and was blown away to hear a Microsoft executive suggest the company is doing about everything in silicon short of setting up a fab on the Redmond campus these days. Apparently, the Windows giant is seeing more and more hardware areas where it wants to play, in addition to the Xbox, Zune and its keyboard and mouse business. With more than its share of the profits of the computer industry at its command, it has the wherewithal to go after new hardware opportunities.

Thus Microsoft is now co-developing some form of media chip with a fables company because it could not get what it wanted in the merchant market. It is also investigating silicon opportunities in digital cameras, even as it helps drive the next generation JPEG XR standard forward.

In a sign of how serious the Microsoft silicon drive is the exec said "we are looking for [chip-level] tools to handle global resource contention and various kinds of bus and core interconnect architectures" for asymmetric embedded multi-core processors.

Microsoft may not be as far down the curve on system-on-chip design as ARM and the rest of the semiconductor industry, but I don't doubt they have the bucks, people and guts to drive fresh directions. The company's work with the Berkeley RAMP system may be a bellwether for things to come. Stay tuned.

Tuesday, October 02, 2007

Ultrawideband narrowing?

I hear at least one prominent ultrawideband startup is on shaky ground these days. Layoffs look likely and a buy out may be in the works, I am told. It's an unfortunate situation because the technology has plenty of promise, especially in home networking.

It's not that surprising that UWB may be in a consolidation phase. Plenty of startups have grown up around this emerging technology over the past five years, seeded by the usual more than adequate supply of competing venture capital bucks. But there have been plenty of technical and regulatory hurdles to getting products out the door. Indeed, last year Moto and Freescale backed off UWB after making a big push in the area. The wireless USB standard has taken a year or more longer to come to market than backers expected. And consumer electronics OEMs and especially service providers—among the biggest targets for UWB—are pretty conservative when it comes to sending high quality media wirelessly.

Hopefully markets like wireless USB and some proprietary personal area nets will take off soon to give the many fledglings some traction. But in the meantime I am seeking input from anyone who knows about the startup (or maybe there is more than one) in trouble. Post your comment or drop a line to

Monday, October 01, 2007

NAND drive rides PCIe

How soon they forget! As far as I can tell I was the first tech journalist to write about Fusion I/O (Salt Lake City) and their plans for a novel controller to take solid-state drives to a new level. Last week the startup formally launched its first product, without bothering to drop me a line.

The ioDrive is a 4x PCIe card for servers that delivers 80-640 Gbytes. I think it uses multi-level cell flash which typically has a reduced operating life over the traditional single-cell chips, but the company's Web site is vague. It is clear on performance, stating the drive executes 100,000 IOPS (input/output per second) per card with sustained data rates of 800MB/sec (read) and 600MB/sec (write).

That compares favorably with high-end, solid-state server drives such as the 3.5-inch Zeus from Stec that packs 146 Gbytes and delivers 50,000 I/OPS riding hard disk interconnects such as SATA. Fusion IO showed its card running in a Hewlett-Packard blade server at last week's Demo Fall '07 conference, but it has not detailed the technology inside its controller chip yet. (I've got a call into them.)

The company does say the controller uses error correction and wear-leveling to deliver a service life of eight years compared to the five year service life of hard drives. The cards, which will be available early next year for $30 per Gbyte, can be used for either local storage or CPU caching across a wide range of server apps.

Although Fusion IO seems to have an edge, this is an increasingly crowded space. I have heard there are a few other companies with flash controllers in the works, including Marvell. And Intel's Pat Gelsinger held up a solid-stat drive at the Intel Developer Forum just two weeks ago and promised the world's biggest chip maker will get into this market, too.

What will Intel do? Will it make a difference? Heard any details on upcoming controllers from Marvell or startups? Drop a comment or email me at

Wednesday, September 26, 2007

Parting thoughts on

A few random observations today coming back to the Valley from the conference in Austin:

The group is ramping up an effort to set common standards for debug across Power chips and tools. IBM exec Chris Ng who heads the effort is reaching out to industry for feedback. Right now his group is discussing adopting the Aurora interconnect from Xilinx.

IBM's Hollis Blanchard is starting a separate effort in to define virtualization for Power, especially in the embedded world. Among its needs are virtual I/O support in chip sets, the kind of thing the PCI SIG is developing for x86 computing.

To make this all run right we will need not only new hooks in the processors and memory management units but in the embedded OSes, too. So far the RTOS crowd is keeping mum on any plans to support virtualization, Blanchard reports.

The group has at least half a dozen other initiatives in the works, many of them led by resourceful IBM middle managers. Whether the efforts will drive the Power community forward remains to be seen, but there is clearly no lack of effort.

Finally, a quick welcome back to Tracy Richardson. The former chief executive of StarGen is now heading up marketing at interconnect specialist Tundra Semi. Tundra is cooking up a concept it calls intelligent I/O, but it's still in stealth mode, he told me at the event.

Tuesday, September 25, 2007

Power's bus woes

I thought Intel had on-chip bus troubles because it chose to get stuck in the arcane world of PCI bus semantics. That's a happy old jalopy compared to the accident waiting to happen over at Power city.

The four chip makers of the Power world—AMCC, Freescale, IBM and PA Semi—all preach the silicon religion of the day. We're headed for a future of heterogenous multi-core chips. They'd like to ride a common on-chip bus to get there, but I don't see how it's going to happen.

Freescale created its CoreNet on-chip bus as part of the multi-core architecture it debuted in June that will take it from a family of 45nm many core parts next year into a 32-core future someday. For now, Freescale is keeping a lid on the specs of the bus, not even sharing them with Power partner IBM Corp.

IBM, which carries a lot of the water for the Power group, had been plowing the way for on-chip buses with its PLB-4 aka Core Connect. Just when or whether IBM pushes that to a PLB-6 is still unclear. Meanwhile, AMCC which lacks the resources of its two bigger partners has licensed the ARM AXI bus, and startup PA Semi uses its own proprietary on-chip bus.

The group which supposedly oversees the Power architecture—all decisions are made by a separate committee of IBM and Freescale engineers—says it is investigating a 10-50 Gbyte/second high-end on-chip bus. Don't hold your breath. For today, everyone in this group is going their own separate way into the multi-core future, an approach this already fragmented architecture can hardly afford.

Sunday, September 23, 2007

A look inside the x86

As usual, last week's Intel Developer Forum was a fire hose of information with several useful droplets here and there buried in a huge flow of generally useless hype and executives who were both amazingly available to the press and well trained about not saying anything substantive to them. The big interconnect bits included the news about USB 3.0, wireless USB 1.1 and PCI Express 3.0 I have blogged about below.

There were two other interconnect tidbits somewhat buried in my EE Times wrap up in print this week. Intel gave a marketing name to what it had been calling CSI. However, the company gave almost no detail about the QuickPath coherent interconnect to be used starting with its Nehalem processors late next year, except that it will gluelessly support two and four-way servers. Analysts think it will be more than just Intel does HyperTransport, four years late. But other than the unauthorized but well detailed white paper from fellow journalist David Kanter, little is publicly known.

Separately, it was interesting to hear from Ajay Bhatt that Intel uses PCI software semantics in its proprietary on-chip interconnects. This is probably one of several Achilles Heels that will make it hard for the world's largest software company to step graceful and late into the world of system-on-chip design. This ain't the stepwise world of the 986 running at 6 GHz anymore. Good luck!

Wednesday, September 19, 2007

On USB 3.0 and PCIe 3.0

On Day 2 at IDF in San Francisco, I discovered an interesting little detail of the proposed USB 3.0. A chart (right) displayed on the exhibit floor showed the cabling has not only the two separate lines specified by USB 3.0 but another line dedicated to any USB 2.0 traffic.

I guess that means the cables are three times as expensive as today's cables, I told the engineering manager from Foxconn manning the booth. He assured me there is work afoot to keep costs down despite this and the fact there is an extra set of mechanical links in the connector as well. He also said the spec is pretty far along at this point, something the diagrams made clear.

Separately, I listened to a chalk talk by Ajay Bhatt, chief I/O architect and Intel fellow, who filled in some missing details about Intel's Geneseo proposal. Turns out this was actually Intel's proposal for PCI Express 3.0 as made to the PCI SIG. Intel could not reveal that fact last year because the proposals were under NDA, so it released the details as a proposal for an industry interconnect to accelerators. Hmmmmm.

Turns out Intel really likes PCI software semantics as a unifying and simplifying element. It even uses them as a way to describe internal USB and other silicon blocks inside its chip sets and processors linked over its proprietary on-chip interconnects.

How quick is QuickPath?

Thanks to David Kanter for unearthing and sharing more details on Intel's next processor bus than the chip company is ready to disclose. Basically Intel was only willing to share the name—QuickPath—the fact that it connects up to four processors and will debut in Nehalem chips next fall.

Kanter concludes from an analysis of Intel's patents QuickPath will be a 20-bit wide bus and run at 4.8-6.4 GHz, faster than what AMD has planned for HyperTransport in that same timeframe. As Kanter sees it, AMD is going to have a hard time competing with Intel starting next fall when Nehalem debuts through to when AMD gets all its server CPUs revved up on HyperTransport 3.0 by the fall of 2009.

But let's not kick them while they are down, Kanter said. So I note that Tarari chief architect Eric Lemoine said he thinks HyperTransport is a superior architecture to Geneseo because it sports lower latency and provides good access to cache lines. OK, I said something nice for the poor underdog of the PC industry.

Attack of the giant USB

Pat Gelsinger says an amazing 6.2 billion USB devices have shipped to date, more than 2 billion of them just last year. So when Intel says it will ratchet USB 3.0 up to 4.8 GHz, give it new flow control mechanisms and make it support copper and optical—you gotta say look out Firewire. The Intel folk even tested their nearly 0.75 spec at 25 Gbits/s in a software simulation and aim for a protocol that could hit 100G someday.

The reality check, as NEC told me, is that USB 3.0 may need to cut in half its five meter reach to hit its goals and silicon and software will require a major upgrade. Meanwhile, Intel is trying to push wireless USB to 1 Gbit/s before there is hardly any certified WiMeda products for the 480 Mbit/s products, that really only get to about 40 Mbits/s.

That's Intel for you. Lots of ambition and as much execution as it can muster from such a broad industry.

Monday, September 17, 2007

Dispatches from IDF: USB 3.0 debuts

Why, oh why did I wait until the night before to troll through the sessions for this week's Intel Developer Forum. Oi Vay! So now I find out Intel is ready to talk turkey about USB 3.0 on Tuesday.

Last year one of my key PCI Express contacts said there was a lot of work happening in the background on USB 3.0, making use of the 5 GTransfers/second the Express 2.0 spec now enables. Had I only checked back a few weeks ago I might have had a scoop! Now I'll have to elbow my way to the news with the rest of the IDF crowd.

The IDF Web site just talks about a fast synch-and-go interconnect and promises a technical overview and timeline for the technology to come to market. Sounds like it is pretty well thought through at this point. Will we see Gbit/s wired links in 2009? Stay tuned…or post a scoop ASAP as a comment if you know about this stuff.

Also on tap at IDF this week: More details about Geneseo, a collect of interconnect technologies enabling co-processing over Express and the Intel front-side bus. This is Intel's answer to HyperTransport and AMD's Torenza program.

In addition, IDF is host to several discussions of 10Gbit Ethernet and data center interconnects including something under the marketing banner of Intel's Virtualization Technologies for Directed I/O. Whadeva that means!

I hope I get to find out more about all this in between a half dozen sessions touting Intel's upcoming 45nm Penryn processors which Intel hopes will grab all the headlines tomorrow, leaving AMD's latest quad and three-core 65nm CPUs in the dust.

Party time for PCIe

This week's Intel Developer Forum may act as something of a coming out part for PCI Express 2.0. PLX and Mellanox are both touting new Express Gen2 products today and I expect to see news from others when I troll the IDF floor this week. Pericom fired the started gun with a pre-announcement of their chips a few weeks ago, but now multiple companies have real, tested and shipping silicon.

Intel has already announced one chip set for Express 2.0, and I expect to hear about more members of the family this week. AMD and NVidia will likely follow suit to talk about their chip sets expected to emerge in systems before the end of the year.

Computer graphics and muscular PC servers are plowing the way on this change. Admittedly the data path and backplane silicon on routers, switches and other high end comms and military gear hit 6.25 Gbits/second and beyond a while back and they are gearing up for 10G now. But this week's news is the great bulk of the volume computing world is driving these fast interfaces into mainstream electronics.

True, the biggest signal integrity design challenges don't really start to kick in until you get a bit north of the max 5 GTransfers/second Express 2.0 enables. But the PC crowd sees that next step forward and has started real lab work on the migration into that zone too with Express 3.0.

Gentlemen and women, start your engines.

Friday, September 14, 2007

Hot little WiMax handsets?

That's the implication I got from a teaser for a Sept. 26 Motorola press conference at WiMax World in Chicago. The news blast is titled: 'WiMax: Broadband in the palm of your hand,' and includes presentations from Moto, Sprint and Rogers execs.

My read on this sector is it is still way too early for WiMax silicon that fits into the power consumption envelope of a handheld. My gut tells me we are at least 2-3 years away from that day. Perhaps this is just a tease for a pretty boring review of what's happening in WiMax, I don't know.

Once again, I reach out to you, dear readers. If you know what Moto is up to here or have any other news tips coming down from WiMax World, let me know. I will be in Austin that week and will miss the event so I'd like to hear anything you pick up as a comment here or an email to

Thursday, September 13, 2007

Medical's wireless squeeze

You don't find many requirements tighter than these. Power consumption at 100 nanoW and duty cycles of operation just 0.01 percent of the time. That's the goal for wireless links in cardiac implants that aim to stay active 7-10 years before a surgical replacement is needed.

Thanks to Paul Stadnik, RF engineering manager at implant maker Biotronik, for passing on to me a milestone standards makers hit this summer for this technology. The TG30 group responsible for medical standards under the European Telecommunications Standards Institute ratified in July a new low power, low duty cycle protocol for the Medical Implant Communications Services band blessing this use in the 402-405 MHz MICS band.

The standard is used to let implants communicate status information typical once a day with a home monitoring system. The FCC is studying a similar standard. To date it only recognizes this use in the 401 and 406 MHz MICS sidebands, but a ruling aligning US rules with those in Europe could come any day.

Biotronik fielded some of the first implants to use MICS based on its own ASICs. But a handful of chip makers are delivering MICS chips now including AMI Semi, Texas Instruments and Zarlink.

Meanwhile TG30 has started a two-year effort to identify new spectrum needs for implants that need more than the 100 Kbits/s or so you can get with MICS and fancy modulation schemes. Brain implants that capture signals and transmit them wirelessly to controllers for artificial limbs are one of the chief applications for a broadband implant standard.