I checked the SiRF website and noted their three announcements for Nokia, Hitachi, and Intel.
Re: Nokia
This seems to be the most important announcement as it indicates a deep-pockets partner willing to invest $3M to take a look at SiRF's technology and business plans. With a strong link to Nokia, SiRF could have an easy design-in for E911 on Nokia phones. Still there is Motorola, Qualcomm, etc, so there is a big chunk of the cellphone market that is still available (though I imagine Motorola would go with their own GPS).
There has been about $20M of venture capital put into SiRF so far with a recent funding round of $10M completed this July, so $3M from Nokia isn't likely to make a huge difference to SiRF nor does it necessarily indicate interest by Nokia to the exclusion of all other possibilities. I've found no info on SiRF sales/revenues to date.
Re: Hitachi
Creating a reference design for their GPS chipsets enabling easy design-in with Hitachi embedded processors is a plus (and speculatively may be why Nokia got interested in SiRF- does anyone know if Nokia cellphones use Hitachi processors?). But I don't see this as a major advantage nor that hard to duplicate.
Re: Intel
The WinSirf model apparently saves the signal processor (using the Pentium to perform those functions) leaving mostly the RF front-end, to reduce GPS chip-set cost. But as Pentiums are premium-priced versus dedicated embedded signal processors, a designer would have to see the total system cost breakdown to see if this architecture is cost-justified. I've seen benchmarks of Pentium MMX's as signal processors and they can be very competitive with TI, Motorola and Analog Devices DSP engines in performance. The problem is that the Pentium MMXs costs $150-500 while the DSP engines cost $10-20. Using up too many Pentium CPU cycles may mean increased system cost that outweighs the other savings. For occasional applications, it may make sense, but for dedicated/continuous applications (such as GPS in AutoPC), it may not. Also, not necessarily that hard to duplicate if the model is successful somehow.
Overall
In my opinion, the last two appear to be name dropping or basic co-marketing type of arrangements. I didn't read of the kind of solid commitment/investment from Hitachi or Intel that would necessarily bring SiRF a lot more business. In any case, the real technology/cost win would probably be a single chip mixed mode ASIC with RF front-end, signal processor, and basic processor integrated together, which no one has got yet.
However, the Nokia announcement *is* more indicative of a relationship that could be very strong. Something to watch.
FWIW, the Chairman of SiRF was also the Chairman of Acclaim Networks, that just got bought out by Level One, and was a founder of Chips & Technology and S3. Both Acclaim and SiRF were/are heavily staffed by S3 expatriates. These are cases of start-ups developing custom ASIC's to sell in low-margin, high volume "commodity" applications. The direction of SiRF seems to be in this direction also. As long as TRMB focuses on unique high-margin apps with lots of IP protection, I think they are better off staying out of the jelly-bean business.
|
