Whether you design or purchase the CPU board to power your next embedded project, the processor architecture and integrated features are important elements in the selection process. This architecture will become the heart of the new embedded system and will influence system performance, component count, future updates, I/O configuration, and overall power dissipation. Fueled by recent announcements of new silicon architectures targeting long-life embedded applications, multicore processors have moved into the spotlight and now offer designers a wide range of advantages over their single-core counterparts.

One of the prime reasons to incorporate multicore is to boost performance through parallel processing. Developers have access to multiple techniques to enable this performance gain, including symmetric or asymmetric multiprocessing and virtualization. In the symmetric case, a single operating system allocates threads or tasks across the available cores while managing common memory and hardware resources. In contrast, asymmetric multiprocessing allows each core to run independent software so that a single system can easily combine real-time, deterministic tasks with a graphical user interface. With virtualization, a hypervisor isolates and allocates system resources between the operating environments so that real-time, general-purpose, and legacy software can be readily integrated in a multicore system.

Multicore shrinks embedded designs

By combining several functions into a single package, multicore-based systems often result in fewer overall components and lower recurring costs. For example, the latest embedded multicore offerings combine a graphics processor with one or more general-purpose processors. This not only eliminates the separate display processor, but also simplifies and speeds up graphics processing by sharing internal cache and memory with the CPU cores. Designers can also use additional cores to lower the component count by integrating external DSP or FPGA units dedicated to signal processing or specialty functions. Along with this reduced package count, multicore systems benefit from lower power dissipation and smaller form factors. Multicore embedded system designers may be able to reduce or eliminate the cooling fan, while portable devices can benefit from both a smaller battery and enclosure.

At this year’s Consumer Electronics Show, Intel announced its 2nd Generation Core family based on 32nm process technology, which included seven multicore processors that support extended life cycle embedded applications. These Core i3/i5/i7 processors combine either two or four CPU cores, an integrated graphics processor, Last Level Cache (LLC), and a system agent/memory controller that all communicate using a scalable on-die ring interconnect system. The ring technology allows Intel to adjust the number of cores and create variations that optimize cost, performance, and power requirements for a wide range of embedded applications. Designers can select a version that not only covers current requirements, but also leaves room for future performance enhancements.

Performance boost for intense workloads

The Intel 2nd Generation Core processors also include a number of new or expanded technologies to enhance embedded designs. For example, all of the CPU cores (including the graphics core) feature Intel Turbo Boost Technology, allowing clock frequencies to scale up temporarily. The processors also include a new 256-bit instruction set called Advanced Vector Extensions (AVX), which is backward compatible with previous x86 ISA extensions and is optimized for vector and scalar data sets such as those found in embedded signal processing applications.

Targeting the Industrial Platform Computing (IPC) market segment of the embedded industry, MSI recently upgraded its IM-QM67 motherboard to accept the 2nd Generation Intel Core processor family (see Figure 1). The new IM-QM67, based on the Intel Core i7 and Core i5 processors and the Intel QM67 Express chipset, includes multiple display outputs with LVDS, VGA, DVI, HDMI, and embedded DisplayPort interfaces. The board also features Direct-X 10 Shader Model 4.0 and full hardware acceleration, delivering 3D graphics and support for up to 1080P high-definition video. Intel Active Management Technology provides the user with certificate-based activation, reconfiguration, and possibly, deactivation of remote embedded systems, regardless of the current operational status. This Mini-ITX form factor board features dual SO-DIMM slots for up to 8 GB of RAM, one CompactFlash slot, and five SATA ports for data storage along with one PCI slot for I/O expansion.

AMD also announced a new family of multicore platforms aimed at the embedded market earlier this year. The AMD Fusion Embedded G-Series processor is based on the “Bobcat” architecture. It combines low-power x86 processor cores and a DirectX 11-capable graphics processing unit into a single Accelerated Processing Unit (APU) for embedded applications. APU configurations are available with single- or dual-processor cores, at 9 W or 18 W Thermal Design Power (TDP), and at two levels of graphics and video performance. The graphics unit includes multiple configurable parallel processing units that can also be used for compute-intensive, non-graphics processing tasks such as encryption/decryption and network packet processing.

Each APU drives one or two high-resolution displays with built-in hardware decode support for H.264, VC-1, MPEG2, WMV, DivX, and Adobe Flash. The APU is also compatible with both OpenCL and DirectCompute APIs, allowing developers to create multi-thread, parallel software analysis functions for real-time pattern recognition in applications such as video surveillance, radar data analysis, and medical imaging. Designers can choose from multiple I/O controller hub options, available for the APU depending on the desired feature set. For example, the A50M chipset supplies 6 Gbps SATA, Generation 2 PCI Express, and HD Audio. For high end applications, the A55E I/O controller hub includes Gigabit Ethernet MAC, RAID support with FIS-based switching, and a PCI Local bus.

Multicore module fits graphics projects

Shortly after the introduction of the AMD Fusion Embedded G-Series processor, several manufacturers announced compatible, off-the-shelf modules based on popular embedded standards. For example, congatec revealed a new COM Express module, conga-BAF, to target low-power, graphics-based embedded projects (see Figure 2). The module was designed around five of the AMD Embedded G-Series processors, ranging from the AMD T44R 1.2 GHz single-core processor with 9 W TDP, to the AMD T56N 1.6 GHz dual-core processor with 18 W TDP. The conga-BAF has two DIMM slots for up to 8 GB of DDR3 RAM and allows dual independent displays with interfaces for VGA, LVDS, DisplayPort, and DVI/HDMI. Additional I/O options include six PCI Express x1 lanes, eight USB 2.0 ports, four SATA ports, one EIDE, and a 1 Gigabit Ethernet interface. Conforming to the COM Express basic form factor (4.9 x 3.7 inches), the module operates over the 32 ºF to 140 ºF temperature range. It supports multiple operating systems including Windows 7, Windows XP, Windows Embedded Standard, Linux, and QNX.

These new Intel and AMD multicore processors just offered to designers are closely tuned to the evolving expectations of the latest embedded projects. In addition to the promise of long term availability, the processors feature increased performance and lower power while integrating the graphics processor to simplify and enhance the user interface. Both of the new multicore families are based on the x86 architecture, allowing embedded system designers to lower development costs and schedules by integrating widely available personal computer software. Multiple applications such as digital signage, set-top boxes, information kiosks, point-of-sale machines, and gaming platforms can take advantage of some of the same low-cost, off-the-shelf hardware components and software tools used for desktop development. Regardless of the application, designers should investigate the latest multicore processors on any new development project, and as a result, customers can expect to see a new era of smaller, faster, and lower-priced embedded products.

Warren Webb’s background includes more than 30 years as an engineer and entrepreneur developing high-tech products for the aerospace and healthcare industries. Most recently, he has been writing articles on hardware design, software development, and emerging technologies for international trade magazines. Warren holds an MBA from Pepperdine University, an MS in Electrical Engineering from San Diego State, and a High Honors BSEE from the University of Tennessee.

OpenSystems Media [email protected] www.opensystemsmedia.com