When the GNU project was started, the knowledge required to create
something like Linux was rather hard core. But now a lot of people know
their way around, and it doesn't seem so esoteric any more.
The central people working on Linux still have a great deal of valuable
knowledge and experience that the rest of us respect, but the basic
information needed to join in is becoming more and more accessible.
Today, fast custom hardware is expensive to produce. One problem is the
costs of producing multi-layer PCBs, chips in factories etc. But in my
opinion, the most significant problem is: the knowledge required to
build something like a high-speed computer motherboard, or a modern
processor, is still rather hard core. The need for an expensive, hard
core infrastructure follows from this. Try building a 400MHz digital
circuit sometime, you will know what I mean.
I think that will change in time. Manufacturing of devices will become
cheaper and more accessible as "roll your own" technology is developed.
The knowledge required will become less esoteric, and also more widely
known. In short, you'll be able to throw something together in your
living room that requires a whole cooperating industry today.
For now, try FPGAs. They're large enough to hold small processors and
motherboard glue logic (tie a few together) and fast enough for things
like PCI and Gigabit Ethernet (barely). Prices vary.
Sadly all the useful mainstream architectures are not publically
documented, so there's no way to write a complete Free (GNU semantics)
tool chain. However you can target the input that the FPGA vendor tools
accept, many of which are free with limitations.
I look forward to the day someone writes a hardware backend for GCC.
But most exciting for me is the day "roll your own" hardware technology
is readily accessible to me. With a good manual :-)
-- Jamie
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/