On the heels of its rollout of the highly praised S5PC100 application processor in the new iPhone 3GS, Samsung's System LSI Division has just announced a “souped-up” version of its ARM Cortex-A8 implementation, code-named “Hummingbird.” Based on IP licensed from Intrinsity Inc. (Austin, Texas), the Hummingbird is fabricated in 45nm low-leakage CMOS and is rated at 2000 DMIPS and 1 GHz at 1.2 V. The new core is a formidable competitor in future portable applications (like maybe a future iPhone 3GS+ running HSPA+?).

Although Samsung's first implementation of the Cortex-A8 is the aforementioned S5PC100 application processor, it is ironic that the other Samsung, the cell phone company, recently launched a Smartphone in Europe (the i8910HD, sold through Orange) based on TI's OMAP 3430 implementation of the Cortex-A8. As Samsung's LSI operation is rated the No. 2 supplier in the $2 billion cell phone application processor market, trailing only Texas Instruments, these are interesting times.

Freescale abandons selling cellular division

Unable to get an attractive price for its Cellular Products Division, Freescale has decided to continue its cellular operations, but on a smaller level, through supporting its current Integrated Digital Enhanced Network (iDEN) business with Motorola and RIM. Staff reductions have brought the division down to a more appropriate size. Freescale's cellular revenues in Q2 were up 45 percent to the $138 million level, attributable to Sprint Nextel's promotion of its flat-rate "Boost" (iDEN) service. RIM serves the active Latin American iDEN market. iDEN is currently the only packet-based cellular network (but using frame relay, not Internet Protocol), popularized early on by Nextel for its push-to-talk capabilities. However, long-range prospects are not good. This market will continue to decline as 3G continues its worldwide build-out.

Tensilica scores again in LTE

Dresden Germany-based startup Blue Wonder Communications GmbH has licensed Tensilica's Xtensa processor for a new LTE baseband product. I mentioned in previous newsletters that Xtensa has been licensed by three Japanese chip companies for future LTE basebands, so Tensilica appears to have growing LTE momentum. The Dresden company was founded by former executives who founded Systemonic GmbH. Philips acquired Systemonic GmbH and later shut it down as successor NXP’s wireless operations merged with STMicroelectronics. Consequently, an experienced body of knowledge stands behind the startup, and the company demonstrated a prototype software modem at the recent Mobile World Congress. “Blue Wonder” is the popular name of the famous 19th century bridge over the Elbe River in Dresden.

LTE basebands becoming plentiful?

In addition to the LTE baseband developments mentioned above, there are other LTE baseband chips on the horizon...not only from QUALCOMM, ST-Ericsson and Infineon, but also from start-up companies. Altair Semiconductor claims its new LTE baseband chip will be sampling in Q4 of this year, SandBridge Technologies demonstrated its LTE baseband at the last Mobile World Congress, and others including Comsys Communications and NEC have announced (or pre-announced) LTE basebands for 2010. Several unnamed companies are said to be developing LTE basebands based on silicon platforms from CEVA and PicoChip.

As cited in an earlier newsletter, both LG and Samsung are actively developing their own LTE baseband chips, which we believe are more for licensing leverage than for internal use or sales to their competitors. After all, their internal quantities would probably not be high enough to bring the costs down fast enough, and we doubt that Nokia would want to buy from either of them.

And there are those who claim to be actively working on developing TD-LTE basebands, like China-based Datang and Hisilicon Technologies (formerly the ASIC division of Huawei). The time-division duplex (TD) version of LTE is favored for the China market.

The key WiMAX chip houses will no doubt want to migrate their OFDM basebands to LTE functionality. As the president of one WiMAX chip house once told me, “We view WiMAX as ‘training wheels’ for LTE.”

LTE protocol stacks coming from several sources

One really doesn't have an LTE baseband chip unless a significant protocol stack for it meets 3GPP specifications. To that end, software startups are actively developing licensable LTE protocol stacks. Some companies, like MimoOn GmbH, specialize in physical-layer stacks, while others like Cambridge-based Qasara, which was founded by engineers from TTPCom/Motorola, SySDSoft (Cairo, Egypt), and 4M Wireless (Luton, UK) specialize in higher-layer L2 and L3 stacks. Focusing on physical-layer stacks will only get you into the data dongle business. For cell phones, the baseband chip must also handle legacy 3G, EDGE, GPRS, GSM, and/or CDMA/CDMA-1xEV-DO. Just porting existing 3G stacks to a new platform involves several man-years of effort, so deep pockets and/or partnerships will be necessary.

Google’s Chrome OS boosts ARM’s prospects

We grew tired of hearing from Intel that its X86 platform (read Atom, in this case) is the only one that can provide “true” Internet experience and that Windows XP or Vista is necessary for that experience. That's now so much hogwash. Ever since Adobe ported its Flash multimedia platform to Linux, there has been no difference in Internet browsing on an ARM platform versus an Atom. Sure, Microsoft's Office 2007 doesn't run on Linux, but that may be a blessing (I really don’t like Office 2007 as compared to Office 2003). I also use free OpenOffice, which runs on Windows, Linux, and more. It was a lifesaver for me when one of my colleagues sent me a huge (30+ worksheets) Excel spreadsheet to edit and promptly went on vacation. The .xls spreadsheet had a problem and would not open in Office 2007, Office 2003, or even Office XP (I was desperate). OpenOffice opened the Excel spreadsheet with no complaint and saved me two weeks’ delay, and I’m now a believer.

Return of the Generation Gap

Internet access now seems to be a generational thing. The average teenager couldn't care less about Office 2007. He or she simply wants Internet access to Facebook, YouTube, Twitter, IM, and email. Ubuntu Linux has proven to be effective for that purpose, and Google believes that with its lightweight, open-source Chrome OS it can even better support “cloud computing.” Google has stated on a Web post, “The operating systems that browsers run on were designed in a pre-web era. Google promises a new, utopian-era OS in which viruses disappear and long waits for systems to crank up are a thing of the past.” Netbooks, Smartbooks, or Mobile Internet Devices (MIDs) with the new operating system will be available in the second half of 2010. Google also promises a gaggle of new Web-based applications that will run on any browser, even on Windows-based PCs. But, I hope Google (or the suppliers) can also provide drivers for a variety of printers.

Needless to say, this is another shot in the arm for ARM, whose Cortex-A8 (and later A9) platforms will be more in evidence this fall as a spate of Smartbooks, MIDs, and netbooks incorporating them will be rolling out.

Intel-Nokia relationship: Our take

Intel and Nokia announced a long-term relationship “to develop a new class of Intel Architecture-based mobile computing device and chipset architectures which will combine the performance of powerful computers with high-bandwidth mobile broadband communications and ubiquitous Internet connectivity.” That's a mouthful. As part of the deal Intel will acquire a Nokia HSPA/3G modem IP license for use in future products. Some have speculated that Intel will be getting back into the cell phone chip business. We don't think so. “Once bitten, twice shy,” is the proper reading, in our opinion. That said, Intel still feels the need to hop on the fast-moving wireless bandwagon in some way and future notebooks, netbooks, etc. will require HSPA/3G connectivity through data dongles or (in Intel's case) internal basebands. Of course Intel will have to port the HSPA/3G modem software from Texas Instruments’ C55 DSP platform to a yet-to-be-named Intel DSP platform. That sounds easier than it really is. Note, however, it would not be unprecedented for TI to manufacture cellular chips for Intel. TI manufactured Intel's 2G baseband chips for Japan's Pacific Digital Cellular (PDC) network for a few years following Intel's DSP Communications Inc. acquisition in 1999.

If Intel really wanted to get back into the cell phone baseband business (as opposed to the mobility netbook or MID baseband business), it would have bid on Freescale's cellular products division for a complete 3G solution. The purchase price would have been far cheaper than the $1.7 billion that Intel dropped on purchasing DSP Communications Inc. (followed by a $600 million write-off and subsequent $600 million sale to Marvell). However, Intel believes that their upcoming Atom-heritage Moorestown chip will be a worthy applications processor base for Smartphones (that Nokia could learn to love).

On Nokia's part, it wants to expand its wireless business beyond cell phones and to that end, I believe that Intel's experience could usher Nokia more quickly into the MID or netbook market. So, it would not be unreasonable to expect a Nokia-labeled netbook based on Moorestown to hit the market in mid-2010.

Shameless plugs

LTE chips are emphasized in Forward Concepts' new “Cellular Handset & Chip Market ’09” study. This totally revised expanded annual market study is an in-depth (594-page) analysis of the top 53 handset makers and provides 2008 market sizes and vendor market shares by air interface. The study also forecasts handset shipments and subscribers by air interface and global region through 2013. And virtually all cell phone integrated circuits and 2008 vendor market shares for each are covered. Chips are also forecast by type through 2013. We believe that this is the most extensive cellular handset and chip study available. Details are at: www.fwdconcepts.com/cell9

The entire semiconductor market is now driven by DSP technology, since without DSP there would be no digital wireless, no multimedia, no VoIP, and no Internet access of any kind. Our newest in-depth market study covers the entire landscape of the integrated circuit market based on DSP technology, not just DSP chips. “DSP Silicon Strategies ’09” is the most extensive study on the planet of markets based on DSP technology. Many components that were once reported as “DSP chips” are no longer. Rather, they are reported as Systems on Chip (SoC) in categories like ASICs or ASSPs. Details are available at: http://www.fwdconcepts.com/DSP’09.

As always, I invite your comments.

Will Strauss
President and Principal Analyst
Forward Concepts
wis@fwdconcepts.com