20 years people were saying the same thing - "Moore's Law will hit the wall in 10 years (due to physics)". And 10 years ago they were saying it. But as long as there's money on the table, somebody will figure out the lithography.
It happens that I much prefer to read a paper volume, since I can do that in places where an e-book would not survive.
As for Moores law, I thought that it was based on observations of what had already happened. As for needing to create a new part for each application, there is another option, which is to assemble a system from a number of chips,(off-the-shelf-parts)Of course, if you are actually going to produce a billion widgets, then a custom IC might possibly be a better choice, but most of the time the development effort of creating a custom device costs more than the price of using several OTS chips. Plus, then there may be a second source for the parts, and a system may even be repairable.
A changing market, yes, but is it really declining? Back in April, the Association of American Publishers (AAP) released its annual estimate of total book sales in the U.S. In 2009, it was $23.9B, a 1.8% decline from the $24.3B in 2008. Not too bad, considering what a year 2009 was for our own industry and many others! For the past 7 years, the publishing industry's CAGR was 1.1%. Slow growth, but still growth.
The really interesting statistic was E-books, which sold $313M in 2009, and increase of 176.6% from 2008.
I'm guessing most of those E-books were sold by Amazon, but now that everyone has their own E-book reader, I would expect to see B&N and Borders have a higher participation rate.
But for many people, even those who own an E-book reader, an afternoon at the bookstore, perusing actual physical books and enjoying a latte or two is an experience they don't want to give up. I guess the financial realities dictate that there will just be a lot fewer actual bookstores in which to enjoy such an afternoon.
Moore's law is anyway going to hit the wall in coming 10 years. We can not keep reducing the size on a chip without either increase in cost per chip (as EUV, double patterning, e-beam technology or any other technology for the say to write below 20nm is going to be expensive than optical lithography) or sacrifice the speed and functionality (higher leakage current). One way to keep the cost per chip under check is by moving to 450mm wafer fabs but not many companies can invest billions as you also said.
The analogy to the book store B&N does not make sense to me. The decline in the readers will affect both the online stores and superstores in the same way. Semiconductor market is growing quite fast and with more demand in memory, power, lighting, smartphone sectors the competition is not going to go down.
Blog Doing Math in FPGAs Tom Burke 15 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 ...