This generally works much better for a laptop than a PC, since it is much less likely that someone will have changed the configuration. The reality is that most of the time everything is the same, but the OS probably can't make that assumption. Windows 7 actually did a good job of minimizing the time, though. I am painfully reminded of this every time I boot my XP work machine.
I think the solution for zero-time booting should be very simple. Most of these devices do a lot of initialization , recognition of peripherals and loading their drivers into the working ram and so on. A simple technique would be have a battery backup for this working ram so that when you shut down your laptop/PC or whatever it is , it is just in a snooze mode. You press a button and up starts the computer in an instant. The penalty will be the battery power that will be consumed. Another way could be to cut a CD of the working copy of the OS ( after initial installation and configuration). If this CD is present in your CD drive, it just loads it into the working ram and starts. Any time you have changed the peripherals or done something new, you do a normal boot up from the hard disk.
According to Kurzweil we are close to the Singularity where we all become last year's model anyway. The fact that we may be getting hardware that is 100x faster than the most sophisticated biological computer but is still so far short of the capability of that computer means that it's a software problem, not a hardware one. More computational power doesn't mean that we know how to either ask it the right questions or program it to get the answers.
An exascale computer would be approximately 100x the estimated processing power of the Human brain. If the AI problem is cracked after that time the machine will outclass us intellectually to the extent we outclass a mouse.
I would think that mankind would benefit in a very significant way if all our machines: laptops, desktops, PDAs, etc. booted in zero or near zero time. Just think how much time is wasted every day waiting for your machine to be ready? How many leave the machine on 24/7 to avoid the delay in power up? How much power could be saved? I do understand the "need for speed"; raw compute power is very helpful for a number of applications modeling complex systems to name one. It was my experience in the ASIC industry that taught me: local / on-chip processing is better. Every time data moved off/on chip, every time external memory or disk needed to be accessed the speed/power costs were significant. Perhaps, we need to change how we perform processing sooner rather than later? Just a thought.
I totally agree with selinz that in computing power more is always better. making super computers play chess like games against the world chess champions is just the fun part of it. But seriously thinking, we have hardly scratched the surface when it comes to understanding this gigantic universe. It still takes us a year to reach Mars by a space craft and the universe is millions of light years in size. Todays supercomputers have to increase their performance many million times so that those brilliant minds working on their theories can simulate, model the universe of such a gigantic in size and finally get some clues about what is this universe in which we , the solar system is nothing but a tiny spec.
I understand. However, I recently began reading the Enders game series..Just finished Children of the Mind (see Ender's game series in Wiki if not familiar). I'd never read any of the series and after reading the main 6, it's not hard to imagine the usefulness of more computing horsepower. (audiobooks from the library have become my mainstay during my commute) As we move into 3D input and output, the demands will be taxed. Heck, it still takes my $3K laptop way too long to boot up...
If we assume that they solve the hardware problems which will enable it to be built and the software problems which will enable it to be programmed, then what exactly are they going to program it to do? Skynet? The meaning of Life, the Universe, and Everything? Or maybe they will race it against a quantum-based PC. Seriously, at some point we need to figure out if this computing paradigm has reached the point of diminishing returns.
Blog Doing Math in FPGAs Tom Burke 20 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 ...