Portability fueling 32-bit expansion

As suppliers offer 32-bit MCUs at a price/performance profile that fits mid- and low-end embedded applications, system designers are changing architectures to give their products the compelling, competitive product differentiators possible with the gained capacity and horsepower in the 32-bit world. Some real-world applications taking advantage of 32-bit devices are Internet-ready handheld devices, digital cameras and video cameras.

Migrating to a 32-bit architecture offers many advantages for engineers and their products. The change will require some investment and redefinition of the design process, but the advantages of switching from an 8-/16-bit design outweigh the time required for hardware and software modifications-in both the short and long-terms.

Moreover, in considering the move to a 32-bit architecture, designers are likewise evaluating the benefits of switching from proprietary architectures to 32-bit architectures that have a significant market acceptance and a broad user base. Such a transition lets an engineer select among multiple suppliers of processors and tools.

At Oki Semiconductor, we target our ARM-based 32-bit microcontrollers to meet the needs of mid- and low-end product requirements. Compared with 8- and 16-bit devices, these controllers offer four times the performance and lower power consumption for the same price. We also aim to reduce system costs by integrating more on-chip functionality, such as timers, DMA and memory controllers, serial ports, A/D conversion channels, PWM and general-purpose I/Os. For these Oki products, the benefits of switching justify the time and possible costs of migrating to a standard-based 32-bit architecture.

Another advantage of switching to a 32-bit controller is the resultant gain in performance headroom for future migration as product capabilities and requirements expand over time. From a performance perspective, engineers are making the transition to 32-bit MCUs because they can handle more complex instructions, such as multiply-accumulate. They can also take advantage of more-sophisticated CPU architectures that are outfitted with multiple instruction pipelines, caches and conditional instruction execution. And because 32-bit devices tend to use tighter silicon geometries, more transistors can be packed into a package even as power consumption goes down.

Cell phones and digital still and video cameras are examples of applicatio ns that require a more-robust microcontroller for handling complex instruction processing. Today's cell phones now include such features as multiple displays, Web browsing, text messaging, extensive address books with search features and voice dialing. The latest digital still and video cameras that are available on the market also require complex instructions to handle the digitization of still and video images as well as voice. Both of these application examples exceed the capabilities of an 8- or 16-bit MCU.

Processing headroom
A 32-bit microcontroller has more processing headroom than an 8-/16-bit MCU and is ideal for instructions written in high-level programming languages such as C, C++ and Java, which are more structured, much more powerful and less time-consuming to write than programs written in assembly language. As a result, software instructions for a 32-bit microcontroller can perform more-complex tasks while taking less time to develop.

Some of the trade-offs for desi gners making the switch from an 8-/16-bit MCU are in the differences of the software languages being used. Most 8-/16-bit MCUs have little overhead and require software to be written tightly to fit their limited processing capabilities. The result is that much of the software written is in assembly language, a lower-level language that is very efficient but that is also difficult and time-intensive to program. Software written for 16-bit MCUs in assembly is not easily portable to other architectures, which means engineers have to redo their software with each new design, which is not the case for 32-bit architectures.

With the benefits of migration come some design challenges for the engineer. Designers will have to make an initial investment in software and tools for the features currently available in a 32-bit MCU in exchange for performance gain. They will also have to determine how much of their software can be used with the new architecture and how much work will be necessary to port the existin g software on to the new 32-bit architecture.

Those engineers who are working with de facto-standard MCUs will benefit from a support infrastructure in the market that can readily provide them with application software, tools, a real-time operating system and a wide base of engineering talent familiar with the architecture. Those factors will ease the transition to 32 bits.

The landscape of the 32-bit MCU world is expanding. As more design engineers responsible for low- an d mid-performance products move from 8-/16-bit MCUs to 32-bit controllers to give their products a competitive advantage, they are also migrating from a captive, proprietary supplier to suppliers that design with a broadly accepted architectures. This switch provides designers with a solution that can be expanded as product capabilities and requirements expand over time.

Mitchell Le is senior product marketing manager for Oki Semiconductor's microcontroller products, and Bahram Raad is an applications engineer fo r Oki Semiconductor's logic systems ICs (Sunnyvale, Calif.).

See related chart