It's rare that construction outstrips creation. We've always been able to design things we can't buildtall buildings, flying cars, or talking robots that are impossible to construct without some breakthrough in manufacturing. But with embedded chips we can build more than we can design.
Developers have never had it so good. Hardware engineers and chip designers no longer worry about the plebeian details of transistors, bits, current drive, VHDL, and so forth. They now lord over big, large-scale, high levels of abstraction like monarchs commanding armies by pushing scale models across a map. Instead of struggling with technical details, we struggle to expand our imaginations.
It's rare that construction outstrips creation. We've always been able to design things we can't buildtall buildings, flying cars, or talking robots that are impossible to construct without some breakthrough in manufacturing. But with embedded chips we can build more than we can design. Our manufacturing prowess exceeds our design grasp, for perhaps the first time in history.
The fabled (and usually misquoted) Moore's Law throws transistors at us at a rapidly compounding rate, equivalent to about a 58% annual return. Like a fantastically wealthy tycoon, we simply can't think of ways to spend it fast enough. We suffer from an embarrassment of riches. Give a bricklayer 100 bricks and an order for a brick wall, and the job gets done swiftly enough. But give him 100 million bricks and 18 months to use them, and suddenly no one knows what to build.
Obviously, you can't design a skyscraper one brick at a time. And modern chips aren't designed one transistor at a time, nor are large programs created one bit at a time. Yet our ability to harness design complexity hasn't kept up with our ability to manufacture the little darlings. Like all monarchs, we're besieged by supplicants bearing foolproof plans for redistributing the wealth and power. New design tools, new programming languages, and new EDA tools all promise to manage the rich harvest of our silicon fields, while incidentally taking a small pittance for themselves.
These new approaches to programming and chip design are many and varied. Most have sound technical merits, yet none has won favor.
Our industry thrives on innovation and inertia in equal measure. We crave higher productivity but hate adopting new ways or new tools. We favor the status quo over revolutionary change. Meanwhile, phenomenal riches go largely unexploited. It's anyone's guess which new approachif anywill catch on and allow design and imagination to catch up with production. We've never had it so good, but it could be even better.