Get More For Less In This Year's Microcontrollers

Jan. 17, 2008
Smaller. Less power. More multicore. Virtualization. That’s 2008 in a nutshell. The microcontroller, processor, and digital signal controller (DSC) arenas will be rife with new products and technology. 64-Bit

Smaller. Less power. More multicore. Virtualization. That’s 2008 in a nutshell. The microcontroller, processor, and digital signal controller (DSC) arenas will be rife with new products and technology.

64-Bit Virtualization
This year will see mass shipments of Intel’s and AMD’s latest quad-core offerings with their respective virtual-machine management versions. This includes AMD’s quad-core Phenom platform, which should give Intel a run for its money (Fig. 1). The Power architecture will remain a potent force as well.

The multicore aspect is important, but the new support is critical to the use of virtualization. These new chips reduce the overhead associated with virtualization. Part of the issue is that virtualization is one way to partition the use of cores, especially since this approach allows a range of schemes, from simple allocation of a single virtual machine to a core to virtual machines that can float among chips.

The minimal overhead eliminates the excuse to get closer to the iron without protection. Virtualization provides protection as well as flexibility. Also, virtualization support will continue to change, but this year’s concentration will be on I/O virtualization. The chip sets and PCI Express chips will make this real in 2008.

More Cores: Supersize It
Four cores on a chip may seem like a lot in some venues. But designers will have their pick of even higher core densities now that many of last year’s multicore designs will be shipping this year.

The trick for vendors of these higher-count multicore chips is to deliver on the software tool side. The programming of these chips remains specialized, but the payoffs for a particular application can be immense.

Look for more improvements in research efforts like Intel’s Terascale project (Fig. 2). Still, massive core counts need not be found only in the lab. Check out stream processing solutions from the likes of Nvidia and AMD ATI.

These systems are based on existing graphics-processingunit (GPU) architectures that employ hundreds of floatingpoint processors. The chips are available, but it’s more likely that standard PCI Express-based boards will be used initially in systems to augment conventional multicore processors.

Signal Processing
Look for multicore design to continue its creep into the signal-processing realm. High-end, floating-point digital signal processors (DSPs) will continue their trend toward speed and efficiency while the low end will look for power savings.

Low-end, flash-based DSCs and fixed-point DSPs will bring more computational power to motor control and audio processing. Higher-end solutions will continue their push into areas such as software-defined radio, which can use all the processing power available and then some. Look for tighter integration on video input as well. Those low-cost imaging solutions for cell phones and digital cameras will turn up with DSCs and DSPs sitting behind them.

32-Bit Invasion
The onslaught continues as 32-bit solutions keep pressing down on their 8- and 16-bit counterparts. ARM’s Cortex line will fill out and be adopted by standard component vendors that cut their teeth on the ARM7 and ARM9 architectures. Higher-end standard parts are likely to show up as well.

More recent 32-bit architecture solutions like Microchip’s PIC32 and Atmel’s AVR32 are looking to attack existing solutions like Freescale’s ColdFire (Fig. 3). Each will trade on its peripheral complement, including specialized support in areas such as multimedia.

The key to these bitspanning architectures will be the 8/32-bit tool chain designed to give developers a migration path. A few developers will choose to take advantage of this path now, but its long-term availability will reassure both 8- and 16-bit developers while making migration relatively painless.

The link between low and high is still tenuous. But this year’s selection is filling out existing product offerings, providing developers with choices of compatible chips they have come to expect from an 8-, 16-, or 32-bit line.

Tiny Micros
How small can they get? How many pins can be crammed into a 3- by 3-mm chip? These and other answers may emerge this year as 8- and 16-bit microcontrollers continue their downward rush away from the 32-bit invasion. Of course, the 8- and 16-bit microcontroller markets continue to compete with each other as well as their low-end 32-bit brethren.

Smaller architectures still will have an edge over 32-bit offerings when it comes to combining microcontrollers with other devices like the MEMS pressure sensors in Freescale’s MPXY8300, as well as in wireless support such as 802.15.4/ ZigBee (Fig. 4).

Look for more activity surrounding the use of devices providing dedicated functions with one or more central processors managing the system. In many cases, it will be simpler to use a processor to front-end a USB connection while another handles the rest of an application, rather than trying to combine both functions on a single chip.

Secure Components Emerge
The most change will occur in security and encryption support. The push to more networked and wireless connectivity demands that devices be as secure as possible. Secure boot, tamper-resistant circuits, and secure key storage will work their way into standard product lines instead of being used in custom or specialized chips.

Standards remain the big challenge. At the high end, systems that can support software such as SELinux will get more of a workout as software vendors fit these solutions on smaller platforms. In the meantime, hardware vendors will reduce overhead by providing hardware acceleration.

Sponsored Recommendations


To join the conversation, and become an exclusive member of Electronic Design, create an account today!