I’ve been saving on an article that appeared in EETimes for a couple of weeks because it makes some very cogent points about System Realization, even though it’s not written with System Realization, or SoC Realization, or even Silicon Realization in mind. The article is titled “Bridging the gap between custom ASICs and ARM-based MCUs” and it was written by Mark Saunders, a Senior Product Marketing Manager at Cypress Semiconductor. The article is really about the Cypress Semiconductor PSOC (programmable SoC) family, which consists of devices that look a lot like microcontrollers with analog and digital peripheral configurability. Although the article is really about a Cypress product family, it covers many System Realization topics quite well, which is why we’re covering it here at EDA360 Insider.
First, writes Saunders, “…the embedded world is a spectrum of complexity ranging from ASIC to MCU.” He goes on to point out that the development costs for ASICs are higher but the “potential for innovation is almost unbounded.” Then he writes that “MCU projects are almost free in comparison and take months or even weeks to finish, but they are limited by the third party chips they run on.”
Well, not quite.
Roughly half of the development cost of any large System Realization project is now tied up in software development and the development cost of the hardware represents the other half. Using an off-the-shelf microcontroller affects the hardware costs but doesn’t really affect the software-development costs except by making an eval or developmnt board available for early software development. The eval board may or may not affect software development costs, depending on eval board’s design resemblance to the final target system. Nevertheless, I agree with Saunders that the spectrum of System Realization hardware choices is anchored on one end by ASICs (or SoCs) and that the other end is anchored by a huge world populated by literally thousands of microcontrollers from a growing number of microcontroller vendors.
Saunders’ article gets another thing right as well. ARM’s entry into the microcontroller business has had profoundly affected the landscape of the microcontroller market. ARM cores have essentially eradicated 16-bit microcontrollers. Almost all new microcontroller development is with 32-bit processor cores, because they offer the performance and address space modern applications need.
And further, I agree: it truly is all about the apps. Saunders writes: “Put simply, software rules because no one wants to port it.” Then Saunders writes some very accurate words about the world of software development: “…in the MCU world, designers live and breathe design in the software-centric Integrated Development Environments (IDEs) that bring together source editing, project management, compilation tools, and debugging together in a single framework.” Here, Saunders hits the System Realization nail squarely on the head. At the System Realization level, software and system developers will be working in an environment they already know and love (well, perhaps love might be too strong a word, but you get the idea) and any company planning to offer System Realization development tools needs to recognize that situation, and deal with it.
Saunder’s article has one more concept well in hand too. There truly is a spectrum of devices that design teams can use to implement their systems. These devices range from ASICs, to FPGAs, to programmable embedded platform ICs like the Cypress Semiconductor PSOC family and the Xilinx EPP family, to the massive profusion of standard MCUs. Part of System Realization must entail being able to pick the right color in the semiconductor spectrum for the project at hand.