To @rickmannow: Another advantage of the ASIC is lower power dissipation, typically you can optimize ASIC circuitry for particular function in question...but on the flip side many ASIC takes much longer than 6 months and 1 Mil to design...Kris
The main differences between SiBlue and the other FPGA vendors are the power and the price of the parts. I don't think any of the other FPGA vendors can offer parts under $2 and I know you can't get an FPGA with the ultra low power of these devices.
As to the use of ASICs, there is only one advantage of an ASIC over an FPGA, unit price. But at such low unit costs the difference can be justified by the many advantages of an FPGA. If you were talking about a $20 FPGA vs a $10 ASIC, then you might not use an FPGA. But if the difference is spending $1 MIL and six months to develop a $1 ASIC vs spending $100k and two months to develop a $2 FPGA which can be updated to fix bugs and add features, the pain of the added unit cost fades away.
I agree @Robotics Developer, my experience resonates with this statistics...so in 1 or 2 cases in a 100 you go to volume production and might eventually be replaced with an ASIC...but you still get revenue from the remaining 98/99 albeit at smaller numbers...it still should be good biz as margins could be high...Kris
In mobile, the saying "be quick or be dead" is a painful truth.
With the changing landscape of mobile designs and the challenges faced by design engineers to meet area,power,performance having an FPGA targeted to mobile apps complements the system needs very well. The key differentiator for 'mobile' FPGA (assuming it meets all criteria of performance,cost,power) is enabling a feature expansion and meeting time to market pressures.
Here's an example of camera and display sub-system that helps semiconductor vendors get to market fast http://www.synopsysoc.org/onthemove/
Low power FPGAs are a good thing and more power (pun intended) to Silicon Blue for bringing another option to the table. If I remember my numbers regarding FPGA (or just plain old Gate Arrays): 1 in 10 opportunities results in prototype design wins, 1 in 10 prototypes results in production orders and 1 in 10 production orders reach the 100K or more level of volume. It really is a numbers game, so for 100 opportunities there may be 1 or 2 going to production. Just thinking out loud...
I think there is a lot value in developing low power FPGA for mobile apps...yes, you will eventually get displaced by an ASIC if you start selling millions of pieces but until then gross profit margins will be very good, and opportunities will be plentiful, 100 design wins sound reasonable to me...does anyone know how exactly they differ from Xlinx and Altera? Kris
I think there must be a timeline to profitability or they would not continue to get funding. Perhaps the economy has caused that timeline to stretch out a bit. A lot of these tablet designs got scrapped and also if anything was in a Palm device that may have gone away with the HP acquisition. Given RIM's issues I am sure they scrapped some designs as well. That may be a partial explanation as to why you are not seeing products.
If they had 100 designwins in Jan 2010, we would have already seen products in the market by now. To take such another 18M in VC funding shows that they still cant survive based on their revenues itself.
Nearly 18 months ago, Kapil Shankar told me that SiliconBlue had roughly 100 design wins (see url at bottom of post). He showed me evidence that at least some of these wins were for things like e-readers, cameras and other handheld, battery powered products. Some people in the FPGA world whose opinions I respect told me privately that they were skeptical of the number. Clearly, though, SiliconBlue is doing something right to be able to tell a story so compelling as to continue to attract significant funding.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...