More is too more

In 1946 only one programmable electronic computer existed in the entire world. A few years later dozens existed; by the 1960s, hundreds. These computers were still so fearsomely expensive that developers worked hard to minimize the resources their programs consumed.

Though the microprocessor caused the price of compute cycles to plummet, individual processors still cost many dollars. By the 1990s, companies such as Microchip and Zilog were already selling complete microcontrollers for sub-dollar prices. For the first time most embedded applications could cost-effectively exploit partitioning into multiple CPUs. However, few developers actually do this; the non-linear schedule/LoC curve is nearly unknown in embedded circles.

Today's cheap transistor-rich ASICs coupled with tiny 32-bit processors provided as "soft" IP means designers can partition their programs over multiple—indeed many—processors to dramatically accelerate development schedules. Faster product delivery means lower engineering costs. When NRE is properly amortized over production quantities, lower engineering costs translate into lower cost of goods. The customer gets that cool new electronic goody faster and cheaper.

Or, you can continue to build huge monolithic programs running on a single processor, turning a win-win situation into one where everyone loses.esp