Intel Architecture: Meeting the Challenge of Fanless Embedded Designs

Q. What can you tell us about Intel’s new platform for thermally-constrained embedded designs?

The new platform is based on the Intel® Atom™ processor Z5xx series and the Intel® System Controller Hub US15W. It incorporates micro-architectural advances that make it possible for the platform to deliver great performance while keeping thermal design power (TDP) to less than 5 watts¹. We are delivering it in a highly integrated 2-chip solution with an extremely small footprint.

Q. Why is 5 watts TDP significant?

When it comes to TDP, 5 watts is a critical number because it allows the use of small form factor fanless designs. This thermal efficiency make the platform an ideal fit for embedded applications, such as in-vehicle infotainment systems, medical tablets, handheld data terminals and entry level point of sale kiosks, and industrial robotics where fans cannot be used.

By providing a new platform for these and other thermally-constrained applications, we are bringing the performance, scalability and connectivity advantages of Intel architecture to an embedded space where it has never been practical before.

Q. What role does Intel’s 45 nanometer technology play in this power optimized architecture?

This processor is one of several new products based on Intel’s 45 nanometer process technology. When you compare it to our prior 65nm technology, Intel’s 45nm technology provides approximately twice the transistor density and more than a 20 percent improvement in transistor switching speed. Higher density enables a Intel to make components smaller, and faster switching supports robust performance.

At the same time, 45 nanometer gives us more than a five-fold reduction in source drain leakage, a greater than 10-fold reduction in gate oxide leakage and a 30 percent drop in transistor switching power. Taken together, the improvements that result from 45 nanometer technology provide the foundation for significantly improved performance-per-watt.

Q. What else has Intel done to optimize power-efficiency?

The Intel Atom processor includes quite a list of power saving microarchitectural features, including an in-order execution core, the new hardware-controlled C6 enhanced sleep state, a CMOS front side bus, dynamic L2 cache sizing, and split VTT rails further reducing leakage current.

Power efficient instruction execution is the heart of the new microarchitecture. In the Intel Atom processor Z5xx series, macro operation-based in-order execution improves decoding and scheduling efficiency and allows multiple instructions to be combined, so the processor can execute them in a single cycle. By itself, the processor has a TDP of about 2 watts, varies by SKU, in a package measuring 13x14 mm, or about the size of a US dime.

Intel’s Hyper-Threading Technology² will be available in the 1.6GHz Intel Atom Processor Z530. HT Technology provides high compute performance-per-watt and increased responsiveness in multi-tasking environments, enhanced by dynamic per-thread L2 cache allocation. The processor also supports the Intel SSE3 instruction set, which enables software to accelerate data processing in complex arithmetic and video decoding, which can be especially useful in gaming applications.

Q. Are there other architectural advantages?

The Intel® System Controller Hub (SCH) US15W is also power-optimized. It integrates a graphics and memory controller hub (GMCH) and an I/O controller hub in a single chip measuring just 22x22 mm, or about the size of a US quarter. The processor and chipset have a unified memory architecture (UMA).

When combined with low-power integrated graphics processing, HD video decode and dual independent display support, UMA means you do not need to worry about the cost of implementing additional video memory into your embedded designs, which leads to board real estate and bill of materials savings. The chipset’s integrated graphics controller with programmable 3D accelerator maximizes performance-per-watt and is designed to reduce memory accesses and power consumption.

The chip provides dual display pipes, providing simultaneous independent support for Low Voltage Differential Signaling (LVDS) with maximum pixel clock of 112 MHz (18 or 24-bit color) and Serial Digital Video Out (SVDO) with maximum pixel clock of 160 MHz. The SCH includes an Advanced Configuration and Power Interface which allows applications to control system sleep, device power, CPU power and performance states. Expansion interfaces include USB 2.0 and PCI Express*.

Q. What are some of the target application segments for the new platform?

As you can see from our overview, this is a low power, high performance 2-chip Intel architecture platform that integrates graphics and video decode in a small form factor with standardized I/O. Those attributes make the platform very compelling anywhere a designer wants to use software applications, such as multimedia software and Internet applications developed on other Intel architecture platforms, with standard network connectivity options.

We are seeing interest from developers of medical patient monitoring and diagnostic devices, in-vehicle infotainment systems, industrial programmable automation and control systems, point of sale kiosks, and entry-level gaming systems that need to put big performance into a very small space.

Q. What does it mean to bring Intel architecture to these embedded segments?

You are getting robust performance, the low power consumption, integrated graphics/video capability, HD video decode and dual-independent display support in a small footprint, sub-5 watt TDP platform that provides the full Intel architecture experience. In-vehicle infotainment systems are an excellent example of an application space that can really benefit from these capabilities.

MP3 players, smart phones, handheld gaming consoles, GPS navigation devices, cameras, DVD players and various kinds of mobile Internet devices are becoming very popular, and this trend presents huge opportunities for car makers. People want everything these devices deliver, wherever they may go, and that includes the hours they spend in their cars. Drivers and passengers want to be connected, to be informed and productive as well as entertained. That is what we mean by ‘infotainment.’ So what we are targeting is seamless integration between devices, vehicles, content, applications – and people.

This means the car’s head unit, the hub of its network, must have the ability to handle digital content, including content protection mechanisms, and interact with multiple audio/video sources and players. People need to access all of these capabilities through an easy to use interface.

For the in-vehicle infotainment segment, you want to support one or more video displays, such as navigation for the front seat and a DVD player for the kids in the back. You want the ability to connect with wireless wide area networks to access Internet services, and the vehicle’s own local network and personal area networks such as Bluetooth* and Ultra-Wideband. You need the processing performance to run more of the popular client-based and rich Internet applications that consumers want. And you want to be able to cost-effectively incorporate new digital content, now and in the future, including new Internet media standards and evolving digital rights management (DRM) technologies.

Many of these applications and technologies are already appearing on the PC. Now vendors serving the automotive industry can benefit from this same architecture, with the added advantages of optimized thermal performance for passive cooling, a form factor that fits within automotive industry standards, and the flexibility to support automotive I/O such as Controller Area Network (CAN), Media Oriented System Transport (MOST) and other networking standards.

Q. What about industrial applications, including portable devices?

The power optimization of Intel’s new platform also makes it ideal for mission critical, fault tolerant application segments that require convection or conduction cooling while meeting stringent requirements for ruggedness. With the correct thermal design solution you can eliminate the power draw, contaminated airflow, noise, and potential point of failure represented by a fan.

In addition to its application handling, the Intel architecture embedded platform provides a highly reliable solution that is designed to bring many of Intel’s advanced platform technologies to embedded designs. As we touched on earlier, HT Technology will be supported by specific versions of the Intel Atom processor Z5xx series, so the processor can run threaded applications.

The industrial segment also includes test and measurement devices, many of which are portable. The platform is not only suitable for ruggedized applications, its ultra-low power characteristics support extended battery life. Combining the new C6 sleep state and 45 nanometer technology results in longer battery life in other embedded applications like ruggedized medical diagnostic tablets, military radio equipment and smart signs.

One of the most important things to remember is the inherent scalability of Intel architecture – you can create all of these devices using the same common architectural foundation used in your office PCs and servers.

Q. Developers already create many of their most powerful applications for Intel architecture. Do you see similar advantages for embedded segments?

This is one of the central advantages of Intel architecture. Just about all leading Internet applications were designed for Intel architecture-based PCs. And as these applications move into thermally constrained embedded devices categories, developers can use Intel architecture throughout their entire product lines for code compatibility. In addition, Intel architecture supports major operating systems, and software drivers, software development tools and board-level solutions are widely available from Intel and members of the Intel® Embedded and Communications Alliance.

With this power optimized platform, we are taking Intel architecture places where it has never gone before.

¹The TDP specification should be used to design the processor thermal solution. TDP is not the maximum theoretical power the processor can generate.

²Hyper-Threading Technology (HT Technology) requires a computer system with an Intel processor supporting HT Technology, and an HT Technology-enabled chipset, BIOS and operating system.