SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Strategies & Market Trends : Gorilla and King Portfolio Candidates -- Ignore unavailable to you. Want to Upgrade?


To: slacker711 who wrote (48958)11/15/2001 11:23:54 AM
From: carranza2  Respond to of 54805
 
Excellent point, slacker.

Perhaps the delays in getting the UMTS standards stabilized will permit Q's competitors to somehow advance their chip-making capabilities for the forthcoming 3G market. But the lack of erosion of Q's pre-eminence even in the low-end CDMAOne chips suggests that Q will also be the maker to beat in 3G markets.

We'll see.

C2@cabalhasbighatsnocattle.com



To: slacker711 who wrote (48958)11/15/2001 3:58:03 PM
From: Eric L  Read Replies (1) | Respond to of 54805
 
re: QCOM TXN Traditional DSP chips v. The ASIC DSP

<< Qualcomm's competition in the ASIC market.... >>

Before I respond to your questions (in a separate post) I'm wondering if you had seen this April DSP Market Bulletin from Forward Concepts that discusses Traditional DSP chips (TXN et al) used on wireless mobile telephony and The ASIC DSP (QCOM et al).

You might have the distinction clear, between a DSP and an ASIC used in mobile wireless, but it is a little fuzzy to me, so on the chance that it might be fuzzy to one or two others here, I thought I'd post an article that somewhat clears up one aspect of my fuzzies.

The article is followed by Forward Concepts latest forecast for programmable DSP's and an abstract from the article below of mobile wireless DSP shipments by vendor last year (2000).

In their expanded 2.5G/3G world it looks like Qualcomm will slug it out primarily with TI, DSPC (and Intel), and Motorola, and a possibly revitalized Agere.

>> The Cellphone Baseband Chip Market

Will Strauss
Forward Concepts
Dsp Market Bulletin
April 4, 2001

forwardconcepts.com

DSP: The Most Embedded Processor

With the Embedded Systems Conference soon upon us, we must again remind the world that digital signal processors (DSPs) are the most embedded of all processors.

And the largest market for DSP chips is the digital wireless market, with cellular phones as the main driver.

We estimate that the cellular terminal or cellphone market reached some 416 million "sell-through" units in 2000. That number doesn't include chip shipments for work in process or for finished goods inventory at the cellphone vendor, so the chip houses can reasonably claim even higher collective shipments. Moreover, some cellphones contain two DSP-based chips: one for the baseband operation and another for the vocoder function, while most employ a single DSP for both functions. So, actual chip unit shipments can well exceed the number of cellphones shipped.

The ASIC DSP Role


The DSPs serving the cellular market come in two varieties: those from traditional DSP chip houses, led by Texas Instruments, and the less obvious ASIC DSPs based on licensed intellectual property (IP) usually in the form of DSP cores.

We have analyzed the market that includes ASIC DSP cores by Qualcomm CDMA Technology or through licensed DSP cores incorporated in baseband chips from merchant market ASIC chip vendors like LSI Logic, Prairiecomm, and Samsung or from significant cellphone vendors themselves, like Siemens and Philips. In addition, we have concluded that cellphones are also the largest market for licensed DSP IP as well.

Our analysis indicates that traditional DSP chips were in 70% of the cellphones shipped in 2000 (292 million units), while licensed DSP cores are in the other 30% (124 million units).

TI clearly leads the traditional DSPs with an estimated 83% of that segment (242 million units), followed by Motorola with about 10% of that segment (28 million units) and the remaining 7% or so mostly from Analog Devices and Agere/Lucent.

If we count the ASIC sockets based on the intellectual property (IP) involved, DSP Group, Inc. is the clear leader, with its Pine, Oak and Teak cores shipping in an estimated 83 million units for 67% of the licensed-core DSP market, followed by Qualcomm with 32% of that segment (an estimated 40 million chipsets, excluding their base station chip shipments). We estimate that another 1 million handsets based on licensed DSP cores (<1%) were shipped by other companies.

On a market-wide basis, TI leads with DSP chips in well over half (58%) of the cellphones that shipped in 2000, while DSP Group is a substantial number-two, with almost a fifth of the worldwide market (20%).

TI can claim sockets at market leader Nokia as well as at Ericsson, Sony, Sendo, Handspring and even (yes) Motorola along with PDC sockets through Intel's acquisition of the former DSP Communications, Inc. DSP Group is known to have its IP in sockets in a comparable number of companies. Although DSPG does not publish details on its licensees, we believe that Siemens is their top licensee followed by Philips; and through ASIC chip vendors like LSI Logic and Samsung, their IP is also in sockets at Bosch, Denon, Mitsubishi, and Panasonic, among others.

The Future


It is in the ASIC DSP space that Intel has its crosshairs on. Intel is working to develop 3G chipsets, perhaps based on the "Frio" DSP engine jointly developed with Analog Devices--coupled with its XScale RISC core of StrongArm heritage. Intel refers to the overall architecture as PCA (Personal Internet Communications Architecture). In this regard, they are challenging TI's already-shipping OMAP platform, which consists of an ARM9 core and TI's C55x DSP core on the same die. Therefore, we have another ASIC DSP vs. traditional DSP battle brewing.

With the very high prices paid for 3G spectrum by the cellular operators, their payback will heavily depend on providing new functionality and new services for which subscribers will pay extra. This could result in future cellphones with a variety of processing engines (both licensed and internally-developed) that are likely to include DSP cores, RISC cores, and a variety of coprocessors and/or accelerators for GPS, Bluetooth connectivity, MP3 playback, etc. <<

Also:

>> Market Forecast Revised Again

Will Strauss
Forward Concepts
Dsp Market Alert
October 17 2001

forwardconcepts.com

In the aftermath of the September 11th terrorist attack on the World Trade Center, it is now evident that the U.S. economy is headed for a recession. In looking over the semiconductor market overall, we have lowered our projections for the monolithic integrated circuit portion of that market from a 25% decline to a 30% drop for 2001.

Unfortunately, DSP growth continues to be in lock step with the overall semiconductor market "train wreck," and Forward Concepts forecasts is lowering its forecast to a 30% drop in DSP shipments in 2001 to the $4.3 billion level. We have also lowered our 2002 DSP shipment forecast to 32% from our earlier 35% projection. Our revised forecast is presented in the table below:

Programmable DSP Chip Shipments - $Millions, Worldwide


'00      '01      '02      '03      '04      '05       CAGR

$6,142 $4,299 $5,675 $7,548 $9,888 $12,854 16%


One Big Negative


Wireline communications, including telephony infrastructure equipment, has been the major disappointment this year, and will continue to be a drag on DSP sales through the second quarter of 2002, in our opinion. One exception appears to be for IP (Internet Protocol) PBX systems and IP Phones. Although Cisco and 3Com are the market leaders, these DSP-enabled products are tough to pick out of the overall revenue declines reported by the two companies.

Some Bright Spots


The cellphone market inventory glut has pretty much disappeared and revenue growth in this segment was positive in the third quarter for both Motorola and TI. Guidance from both companies is for continued improvement in Q4, thanks to 2G shipments to China and GPRS shipments to Europe.

However, the first half of 2001 was such a cellphone disappointment that we have lowered our forecast of cellphone shipments for the full year to 390 million units, down from the 416 million that shipped last year. Of last year's cellphone shipments, though, it now appears that only about 380 million actually reached customers, compounded by DSP chip shipments that overshot the 416 million mark by another 20 million or so, leading to the inventory problem that was clearly evident by the fourth quarter of last year.

The only other big market (more than 100 million DSPs) that has shown flat to slightly positive growth this year is for hard disk drives. That's because the disk drive manufacturers had their inventory glut last year. Other DSP markets showing positive growth this year are smaller, like digital still cameras (a 17-million unit market this year, growing some 57% over last year) and MP3-type Internet audio (now going into portable CD players, camcorders, digital still cameras and cellphones as well as traditional flash memory players). <<

and ...

>> 2000 Unit Shipments And Unit Market Share By Vendor

Traditional DSP chips were in 70% of the cellphones shipped in 2000 (292 million units), while licensed DSP cores are in the other 30% (124 million units).

                                Million    Market
Units Share


Traditional DSP Chips


TI                              242             83%
Motorola 28 10%
Analog Devices + Agere/Lucent 2 27%
Total 292 100%


Licensed DSP Cores


DSP Group, Inc                   83             67%
Qualcomm" 40 32%
Other 1 1%
Total 416 100%


Combined Traditional DSP Chips & Licensed DSP Cores


TI                              242              60%
DSP Group, Inc 83 20%
Qualcomm" 40 10%
Motorola 28 7%
Analog Devices + Agere/Lucent 2 2%
Other 1 1%
Total 416 100.0%


* excludes base station chips

- Eric -



To: slacker711 who wrote (48958)11/15/2001 5:05:20 PM
From: Eric L  Respond to of 54805
 
re: QCOM - the CDMA ASIC Market ...

<< [QCOM] ... architectural control. ... how long does that control matter? Once the standards have been set and the technology had been around for a while we would expect competitors to enter the market. Curious what you think the explanation is.... >>

I'm not sure that I have a very informed opinion ... just a few guesses.

First their is the matter of proprietary control of both the system architecture, the chip architecture, and the resulting standards, all related to the difference in the development cycle for an open standard and a proprietary open standard.

If you take the largest chip market (GSM) and the 3rd largest (TDMA), all chipmakers are on pretty equal footing. As standards evolve, they are all on equal footing and typically they wait till the standard is finalized (with only minor development) and then they develop their design.

There are some exceptions to this. An example would be Motorola starting development on the first GPRS release, with Nokia and others waiting to start till the completion of the 3rd standards release. Another example would be NEC and Panasonic starting earlier than others on WCDMA. In both cases the developers go through some pain and suffering and burn up more R&D expense, but they also advance their learning curve quicker to improve their competitive posture by being earlier to market.

In Qualcomm's case, they are developing before standardizing, and they accrue the advantage of being well along in systems design before any competitor /licensee. They are always first to market, and that’s a sizeable advantage, particularly when it is coupled with more experience and expertise in their native technology than any of their competitor /licensees.

There are some that say that say that their are some secret Qualcomm ingredients that never really make it into the cookbook (standard) and that the standards are not as detailed as, lets say, those in the GSM world. I have no idea if this is the case or not, but obviously we have seen an issue this year that affected (at least) 3 different manufacturers of cdma chipsets (Nokia, Motorola, Sony). The problem related to the synch channel, and in each case the problems were similar but not exactly the same.

I've looked at a fair amount of documentation on this (high level since I'm just a sales dummy) and it looks to me like the problem arose as a result of some early interpretations made (differently by each) during the early IS-95A stage, that were magnified by the transition between IS-95A and IS-95B, and carried over into the 1xRTT phase 0 standard.

I'm not sure how many of the competitive vendor licensees actually developed or evolved their designs for IS-95B but that was primarily implemented in Korea and Japan (with some rather proprietary implementations in each case) and probably Qualcomm and Samsung where the only manufacturers of chips for these markets.

I remember an article in "America's Network" a year or so ago where Perry LaForge commented on the IS-95 standard being thrown together rather quickly and you might remember that it took the better part of a year to get IS-95B packet data really working. I assume that Qualcomm was very involved in hands on tweaking, again advancing their learning quicker than others.

The fact that we haven't seen more competition from licensees is probably due to a combination of factors that include a fairly limited market (in the case of IS-95B particularly) compared to GSM, limited CDMA expertise on the staff of the competitor/licensee, overall expertise of Qualcomm in development discipline and supply chain management, their control of the standard, their ongoing advancement of the standard adding features and functionality, and the priorities of the potential competitor /licensees based on the larger potential market for GSM GPRS, etc.

Let's face it. Qualcomm is difficult to compete against.

Very few competitors will come out of the gate with early 1xRTT product, and perhaps none with 1xEV-DO (comparatively limited market).

Nokia may be the only one to attempt it. Please notice I said attempt. <g>

Despite that cdma has matured into a sizeable mass market, and 1xRTT is going to be around for awhile, and foster an accelerating replacement market. Over the course of the next year I suspect we'll likely see some competition develop that we haven't seen in the past.

Nokia is a danger, but Samsung would be much more dangerous. Intel a possibility, TI? Maybe. Agere?

That's all I don't know. <g>

- Eric -