Chasecreek Systemhouse writes: > > To design software, all you need is a fully functional computer. > > > > To design hardware, you need to create and test a prototype every once > > in a while. That'll cost you. > > > Your logic doesnt follow. > Why, then, isn't Be (BeOS) still around ? > > Plenty of fully functional computers around at the time -- and Yes, I > know Steve Jobs killed off the Apple clone market. But the problem > was Be could not switch architectures fast enough to survive. > > > True point revealed: functional hardware doesnt go very far without > functional Software.
That is an entirely different point. Creating "complex" functional hardware (for a definition of "complex" that changes over time, as proto boards and cheap FGPAs become more capable) is much more expensive than creating software of comparably "complex" functionality. Tooling up for an ASIC production run using current processes costs tens or hundreds of thousands of dollars -- even if you assume the design is bug-free and there is zero cost until the design has taped out. That is a generous assumption given the absense of free software to perform many pre-tape-out steps and the dependence of that design file on a particular ASIC vendor's cell library. Those tens or hundreds of thousands have to be spent by everyone who wants to actually build the chip. If you just need to do a circuit board design, depending on number of layers and other details, ignoring per-unit costs and again assuming zero design costs, you might get off with paying thousands to low tens of thousands of dollars for a production run. Hardware design has very different and higher third-party costs than software design, and the cost to make and test minor revisions can be a significant fraction of the cost to do the initial build. As long as that is true, free hardware is not possible on the same scale as free software or with many of its benefits. Michael Poole