The provocative title should immediately expose this post as an unabashed attempt to stir up debate. I expect my fellow bloggers to weigh in. It’s my fervent hope others from the community will join the fray as well.
In my last post, Designing for the Post-PC Era, I looked at the shift from desktop PCs to smartphones and tablets as the main driver for leading-edge consumer electronics, and how it has affected electronic design and the DesignCon technical program. A handful of news stories about declining PC sales inspired that line of thought.
Moore’s Law has also been in the news lately, with analysts speculating on how long Intel can continue with a business model driven by keeping pace with Gordon Moore’s 40-year-old prediction. We know that it’s not a law of nature, but rather a
self-fulfilling prophecy sustained by Intel’s R&D investments, and supported by the rest of the semiconductor industry.
Simply defined as the doubling of the number of transistors on a chip roughly every two years, we know, instinctively, that the predicted exponential growth must eventually slow down. We should expect the cost/performance implications of the Law will play a role in the timing of the slowdown. After all, the industry faces both technical and financial challenges to stay on the Moore’s Law trajectory.
As engineers, we need to quickly follow the question of “Can we do this?” with “Should we do this?” when faced with any challenge. In the case of sustaining Moore’s Law, it seems to me the answer to the second question lies in projecting sufficient demand for the kind of performance the next semiconductor process node will enable to justify the cost of getting there.
This brings me back to the diminished role of the PC -- the showcase product for CPU performance -- in the consumer electronics marketplace. Another recent article entitled “System-on-chip technology comes of age,” observed the dominance of highly efficient systems-on-chips (SOCs) and systems-in-packages (SIPs) over high-performance CPU chips in smartphones and tablets.
If it’s true that the power and packaging requirements of this growing product segment are better met with SOCs and SIPs than CPUs, does this tilt the semiconductor landscape enough to slow the march of Moore’s Law?
I like to say the DesignCon community includes both engineers who design chips and engineers who design with chips. This strikes me as the perfect group to take up this question, encompassing, as it does, both expertise on creating chips, and applying them to meet the requirements of next-generation products.
Here’s your opportunity to join a debate worth following. Leave a comment below.