It seems these days that nothing ever gets done fast enough. Just a decade ago, typical design cycles for complex chips or systems were often two to three years or more. Over this past decade, though, those cycles have shrunk tremendously, with designers able to trim what were two-year efforts down to periods of about nine months. But many companies are still dissatisfied with the nine-month delay. They've put new goals in place that will shrink that time down to six months, four months, or even less.
Improvements to both design tools and the availability of intellectual-property libraries have played a key role in slashing design time over the last decade. They'll probably be key to further cutting the cycle as well. How the tools and libraries will accomplish that task seems pretty straightforward. They need continued refinement of the tools to improve their speed and accuracy, and larger IP libraries to provide more choice and thus speed the functional design. Those two avenues pretty much sum up what can be done today. Most EDA tool suppliers and IP developers already have plans to continue down those paths.
The future, though, demands more. Higher levels of abstraction can be used to define more of the system earlier in the design cycle. New tools must be created to allow concept-to-physical implementation with as few steps as possible. Advanced diagnostic and testing tools have to be able to handle complex system solutions. They must be able to check out designs ranging from system-on-a-chip solutions to full systems with multiple boards stuffed with complex chips. By leveraging all of these soon-to-be-available resources, designers will be able to greatly speed their efforts and improve the odds of getting the design right the first time.
But why the rush to shrink design cycles so much? Have product lifetimes really become that short once they hit the shelves? Or is it that we're so cost-conscious about the initial cost of the product that we leave no option for field upgrading, preferring to spin an entirely new version? Granted, consumers and designers like variety. Plus, market competition is fierce. Crafting slight redesigns does allow designers to leverage the lion's share of one design. "Enhanced" versions provide some market differentiation.
I just wonder if we've gotten too good at "painting wheels" blue, red, yellow, etc., rather than developing better alternatives to the wheel. If too much attention is paid to these short-term developments, the long-term projects will suffer. That could lead to industry stagnation.
Although I don't think the industry is anywhere near that point, I often see a lot of duplication of effort. The fierce competition between companies and antitrust rules prevent them from directly working with each other. They'd almost never divide up research to make more efficient use of scarce resources. Can something be done to encourage joint developments? Or should we discourage such projects? I look forward to hearing your thoughts and opinions.
Reprints
