A new breed of microprocessors is delivering unprecedented performance at lower power and clock speed, with greater power efficiency levels. 'Multi-core processor,' as the term suggests, provides more than one execution core for the processor—without increasing semiconductor package size, power consumption, or cost.

Because a specific processor core can be dedicated to a given system task, ability to perform true multitasking becomes a reality. This makes the new technology desirable for high-demand and real-time computing. A growing number of applications, especially space-constrained embedded systems, are poised to take advantage of multi-core benefits. To date, Intel Corp. has introduced Dual-Core and Quad-Core processors, based on innovative Intel® Core™ microarchitecture.

Dazzling advances have marked the progress of microprocessor and semiconductor technologies. While device sizes shrink, unabated rise in computing speed and processing power has followed the path of Moore's Law far longer than most anyone predicted. However, even the latest traditional microprocessors ran into performance and power-consumption limits when faced with high-demand applications. Something new was needed.

That recent breakthrough is ability to place more than one independent execution core (or computing engine) in the processor—creating a multi-core processor on a single silicon die. It promises unprecedented new opportunities for microprocessors in wider applications. Multi-core processors offer numerous benefits—among them true multitasking where different tasks can simultaneously access a separate processor core, thereby improving performance. Intel Corp.—with 35-years of expertise in embedded computing, communications, and networking technologies—has been prominent in developing multi-core processors (MCPs).

Improved reliability, performance, and power management are high on a list of benefits. Particularly for industrial and automation applications, multi-core processors provide a higher level of reliability. For example, when closing a critical control loop, ability to dedicate a specific processor core to that task means higher confidence for completing the desired control action. One or more extra cores available to execute critical tasks enables another aspect of industrial applications—real-time control.

A real-time (RT) task with its appropriate operating system can safely run on a dedicated core, while a non-critical task, say, operator interface (HMI), runs on another core under-general purpose operating system like Microsoft* Windows*. As a result, an unforeseen crash of the HMI function doesn't interrupt the application's RT portion. The dedicated core further ensures that its resources are not diluted by other non-critical tasks.

One measure of RT control response is clock jitter, or time difference between ideal and actual closing of a control loop. It translates to positioning accuracy of machine tools or automation systems. Intel® multi-core processors have demonstrated significant improvement in clock jitter. Multi-core processors also can help consolidate diverse control resources for certain applications. Proper software infrastructure is necessary to integrate the separate functions, according to Edwin Verplanke, Intel® platform solutions architect.

Intel multi-core processors incorporate a microarchitecture that distributes device functions and allows more granular power control at the sub-block level. Ways to cut platform power consumption include control of clock speed and selecting low-power skus, explains Babu Narasimhan, segment market manager at Intel. “Scaling down chip control voltage and shutting down parts of the core not in use leads to power savings,” he says. For example, Intel® Core™2 Duo processor (see products section) takes less than half the power to operate compared to an older Intel single-core processor based on Intel NetBurst® microarchitecture.

Other techniques also play into power savings. Smart circuitry of MCPs allows “clock gating,” or switching off specific parts of logic or circuits when not used by an application, notes Verplanke. “We also can do more with gates than before to reduce power while increasing device functionality.”

An important feature is Intel SpeedStep® technology, a control software that allows us to dynamically scale core voltage and frequency to suit conditions. If there is limited activity on a node, computer or embedded device, Intel SpeedStep technology can take the decision to reprogram core voltage and frequency to save power. “Dynamic voltage control is unique to Intel® processors,” says Verplanke.

Overall, Intel® Core™ microarchitecture offers a more efficient execution path for information flow through the Multicore Processor and the chipset (see “Platform” diagram). Processor-to-memory connection is enhanced. Front side bus of the processor is optimized to support the requirements of the memory controller and the I/O controller, adds Verplanke. “Dynamic Execution” and “Smart Cache” diagrams illustrate two important parts of Intel Core microarchitecture.

Lower core voltage and possibility to reduce operating frequency make a smaller semiconductor die more power efficient. However, the die also is more complex to produce. Intel has mastered the 65 nanometer (nm) process node, which is used to produce all its multi-core products. Intel is ahead of its competitors in leading the transition to 65nm technology, thus enabling multicore processors that are smaller and more efficient. At smaller process nodes, leakage current/power becomes a greater part of total or dynamic power. Control of leakage current even in idle circuits becomes necessary for processor efficiency. “The 65-nm process was designed with lower leakage current in mind,” notes Narasimhan. Besides saving power, turning off idle circuits limits leakage current draw as well.

Another innovative technique to enhance power efficiency at the chip level is called “strained silicon.” Applied in Intel's advanced 65-nm process, it forces silicon atoms apart within the semiconductor material to allow less restricted flow of electrons, resulting in less power consumption and heat generation.

Current Intel multi-core processors are based on Intel Core microarchitecture, fabricated with 65-nanometer technology, and offer 64-bit computing. The product lineup includes:

• Dual-Core Intel® Xeon® processor with two execution cores and up to 3.0 GHz operating frequency. Several other lower core speeds are available;
• Intel® Core™ Duo and Intel Core 2 Duo processors with two cores and up to 2.0 GHz and 2.66 GHz speeds, respectively. (Other lower core speeds available); and
• Quad-Core Intel® Xeon® processor, featuring four execution cores with up to 2.66 GHz frequency (also lower values) and four instructions per clock cycle.

A typical product platform also includes chipset, memory and memory controller, communication buses, interface ports, etc.—with components depending on embedded, desktop, workstation, or server type application. Processors and chipsets come with five to seven year lifecycle support to ensure customers' design stability and protect investments.

Quad-Core was first introduced as a server microprocessor. In April 2007, Intel announced a new embedded version of Quad-Core processor at the Embedded Systems Conference Silicon Valley (San Jose, CA).

On a further positive note, acquisition cost of MCP technology is not an issue. Intel multi-core processors command virtually no cost premium for versions available today—due to fabrication efficiencies and evolving process improvements. “Intel® Core™2 Duo processor has the same or slightly lower price than that commanded by a single-core Intel® Pentium® processor just a few years ago,” states Narasimhan. As future microprocessors with substantially larger number of cores come to market, pricing will likely reflect their higher added-value.

Not surprisingly, compatible software code is needed to realize the benefits of multi-core processors. Many applications based on legacy software used with earlier microprocessors were typically singe-threaded and can't take advantage of what multiple cores offer. Users may not see performance improvement just by changing hardware. However, Intel offers several software tools to ease customers' migration path to multi-core processors.

Many developers want to maximize application performance. Intel® VTune™ analyzer gives developers a view of what's happening as an application runs. It identifies areas that take an inordinate amount of processor time. It also helps identify critical paths in an application where adjustments have maximum benefit. Without Intel VTune analyzer, performance tuning a guessing game.

Other migration tools available include Intel® Thread Checker and Intel® Thread Profiler. Checker finds parallelism in software code, while Profiler analyzes and optimizes performance of resulting threaded code needed for best operation of multi-core processors.

The embedded division within Intel has focused on several target market segments to show customers how to take advantage of Multi-Core. Several open source applications were used to demonstrate methods of software migration to Multi-Core. In the security segment an Open Source Intrusion Detection System called Snort was ported to a multi-core environment, where we demonstrated significant performance increase using a pipelining concept. This avoided the need to rewrite the algorithm and provided an easier method to migrate this application to multi-core. A similar method was used to migrate an Open Source Image rendering application called Amide to Multi-Core. In this case, the data (working set) was smartly distributed over four threads, each assigned to a CPU core. Each thread processed one-fourth of the data, which resulted in an overall performance system improvement.

Initial target market segments for multi-core processors have been in servers, data handling, image/voice processing, wireless communication, and high-end medical imaging applications. Wide application in the control and automation arena has been delayed due to several factors, most of which no longer apply. Early dual-core processors were physically larger, power hungry, and ran hotter, which did not make for a good fit within thermal and space constrains faced by industrial users. In addition, industry's cautious nature to adopt new technology and limited user awareness slowed market penetration.

Intel now offers dual-core processors with power consumption as low as 10 W; MCPs for servers have a wider power range. Ultra-low power products, drawing just 4-5 W are available for mobile and custom applications. Intel sees more mainstream process applications ahead, such as HMI (human-machine interface), data-intensive CAD and machine vision systems, and particularly in CNC production machines. Growing sophistication of industrial controllers could accelerate demand for MCPs. “More intensive processing requirements of industrial controllers and associated sensors and autonomous control at the field level are newer market drivers,” adds Intel's Narasimhan.

While a new breed of microprocessors was the focus here, Intel has much more to offer. It provides customers with modular, scalable computing solutions based on solid planning and long-term product continuum.

Multi-core processors may never reach exclusive market segment share. Single-core processors still have substantial mileage left in many lower-end computing systems. Yet, the coming dominance of multi-core processors is evident. In 2006, Intel shipped more MCPs to the mainstream market than their single-core cousins.

Aided by increased user awareness of what multi-core processor technology can offer, the outlook is wide open in industrial automation and control—as well as in a gamut of other applications.

Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, visit Intel Performance Benchmark Limitations

INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL® PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.

All products, platforms, dates, and figures specified are preliminary based on current expectations, and are subject to change without notice.

Intel, the Intel logo, Intel Core, Intel NetBurst, Intel SpeedStep, Xeon, Pentium, and VTune are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries.

* Other names and brands may be claimed as the property of others.

Copyright © 2007, Intel Corporation. All rights reserved.