IoT & Embedded Technology Blog



What’s in a (Processor) Name?

That which we call a system-on-chip by any other name would still feature broad I/O support, integrated memory, and heterogeneous processing.

by Daniel Mandell | 02/10/2016

Embedded processor taxonomies are evolving with the capabilities of the increasingly integrated and powerful technologies they define. The advent of heterogeneous processing architectures and continued miniaturization of chipsets has fueled the rapidly growing system-on-chip (SoC) market over the past five years at the expense of more traditional processor types like central processing units (CPUs) and graphics processors (GPUs). Microcontrollers (MCUs) continue to be a major part of the embedded processor market with even 8-bit devices still seeing strong use prompted by falling costs with little or no increases to MCU application requirements. As these various embedded processor types each assume more functionality and greater overall performance, the lines between them have blurred causing many to question: What’s the difference and does it matter what we call them?

Embedded SoCs have quickly proliferated on the back of mobile consumer devices such as smartphones and tablets. However, the inherent advantages of consolidated processing and I/O resources in a small package have enabled the technology to also spread through a variety of embedded markets such as automotive and medical devices. Many processor vendors and their marketing teams have adopted the “SoC” moniker to convey that their products are comprehensive processing solutions. Further clouding the delineation between processor types is the expanding use of hardware and software accelerators, security and other specialized coprocessors, and support for video/display ports and interfaces. Identifying an embedded processor type boils down to three main questions:

  1. Does the chip feature a heterogeneous processing multicore architecture (e.g. CPU+GPU, CPU+DSP, etc.)?
  2. Does the chip feature a memory management unit (MMU)?
  3. Does the chip feature on-die RAM?


In the end, does it matter what we call them? Of course! In addition to the technical limitations or boundaries of each processor technology, different processor types have distinct advantages in supporting the gamut of embedded device types and application requirements. Embedded CPUs, for example, are used most in high-performance communications and networking equipment in addition to retail and industrial automation systems. On the other hand, SoCs see more use for automotive subsystems and consumer electronics where component consolidation and flexible processing are vital. While there is some overlap among different embedded processor types in terms of features and application support, the market viability of each type will fluctuate with overall vendor/ecosystem support and end user adoption to address growing application requirements.

So, what’s in a name? That which we call a processor by any other name would… change its meaning.

Happy Valentine’s Day to all of our readers! View the 2017 IoT & Embedded Technology Research Outline to learn more.