Our wish list for what features we want on the convergence device keeps getting longer, while form factor keeps getting smaller. Familiar movie. Unfamiliar outcome.
For the longest time, much of the miniaturization movement had taken advantage of decreasing transistor sizes. But that trend requires lengthy design cycles and prohibitive costs, and is decidedly unsuitable for consumer gadgets that are driving today’s miniaturization market. The need to combine DSPs, RF modules, image sensors and camera modules, GPS receivers/transceivers, MP3 players, memory blocks, etc. on a single tiny chip is a daunting task.
What if they are separate chips? The sheer number of ICs in a single product makes placing them on printed-circuit boards (PCBs) and routing them to be interconnected without shorts in a tight layout almost impossible.
Chips of different functions need to be integrated at the package level. Sometimes, dies need to be stacked on top of each other in a single package. Often, packages need to be stacked. This is a whole different set of technological challenges. [Interested readers can read my piece on SiP or this more technical piece from EETimes by Richard Crisp.]
Innovation is coming from companies like Tessera, Amkor, Intel and IBM, but EDA tool vendors are not quite doing their part yet.
One thing is clear: miniaturization will continue. The semiconductor ecosystem needs to rise up to address the challenges. The volume and innovation drivers for chip demand have both shifted completely to the consumer gadget world.
Steve Jobs and John Rubinstein can only weave their magic if the chip side works in tandem with their visions.
This segment is a part in the series : Trend Radar 2008