In January Intel introduced ten new Intel® Core™ processors for the embedded market.  These new processors—the Intel® Core™ i7, Intel® Core™ i5, and Intel® Core™ i3 families—are the successors to the Intel® Core™ 2 family.  The new processors offer a number of major upgrades and features that benefit embedded applications, including:

 

  • Integration of graphics and memory controller onto the CPU, which reduces the chipset from three devices to two.
  • Significant improvements in performance and performance/power ratios thanks to the 32nm Intel® Nehalem architecture.
  • Embedded-specific features including seven-year availability, BGA packaging, and ECC memory support.
  • Rapid adoption by members of the Intel® Embedded Alliance, giving developers immediate access to a wide range of boards and modules.

 

The new Intel Core family targets applications that need power-efficient performance, including telecom, medical, industrial, military/aerospace, digital signage, gaming, and test and measurement.  In this blog we’ll look at the new features of the new Intel Core family and explain how these features are relevant to these embedded applications.

 

At the highest level, the new Intel Core processors are notable for two firsts:  These are the first mainstream 32nm processors, and the first Intel processors to integrate graphics directly onto the processor.  Together these features contribute much of the Intel Core’s advantages over its predecessors.

 

Let’s start by looking at the increased integration.  As shown in Figure 1, the new Intel Core processors bring features previously located on a separate Memory Controller Hub (MCH) into the main CPU—including the graphics processor (GPU), the memory controller, and the PCI Express* 2.0 controller.  This integration reduces the chipset from three devices to two devices, and reduces the chipset area by 30\%.

 

231i4E47F5675917908F

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. The Intel Core chipsets integrate the graphics and memory controller into the CPU.   The Mobile Intel® QM57 Express chipset shown here is used for all processors except the Intel® Core™ i5-660 and Intel® Core™ i3-540, which use either the Intel® Q57 Express or Intel® 3450 chipsets.

 

The increased integration in the chipset frees up valuable board area.  For high-density applications such as military or medical imaging, this board area can be used to squeeze more functionality into the same space.  In the case of the GE Intelligent Platforms 6U CompactPCI* CT12 or the 6U VME VR12 single board computers, for example, the additional board space allows for the provision of up to two PMC or XMC sites, up to 16 GBytes of Flash memory and a broad range of connectivity (Figure 2).

 

 

232iE5735B51553AEEC8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2. Illustration of Intel Core i7 implemented in a 6U CompactPCI SBC. Image courtesy GE Intelligent Platforms, an Associate member of the Intel Embedded Alliance.

 

“The level of integration within the chip with memory and graphics is a critical” explains Frank Willis, Director of Military/Aerospace Product Management at GE Intelligent Platforms.  “In a 3U or 6U we are scrambling for all the space we can get.  The size of the packaging allows us to get a higher level of functionality on the board than ever before.”  (For more on the CT12/VR12 and on the Intel Core i7 in general, see The significance of Intel’s Intel Core i7 to embedded computing (PDF, registration required).)

 

The reduced board area is also an obvious plus for small-form-factor (SFF) applications.  For these applications, the new Intel Core family makes it possible to shrink the solution into a smaller space.

 

Of course, the integration in the new Intel Core family is a plus only if the graphics performance is good enough to obviate the need for a discrete GPU.  Happily, the Intel Core processors incorporate the new Intel® HD Graphics engine.  This graphics engine is much faster than the engines in previous Intel chipsets, and it compares favorably to competing Integrated Graphics Processors (IGPs).  In my judgment, the Intel Core graphics capabilities are more than enough for the vast majority of embedded applications.  (We’ll look more closely at the performance of the Intel HD Graphics engine in a future blog.)

 

The other headline-grabbing upgrade is the process shrink to 32nm.  This is a major advantage for the Intel Core family, as other embedded processors are currently fabricated in 45nm or larger geometries.  The process advantage helps the new family outperform its predecessors while keeping power consumption down.   The Intel Core family also gains a speed and efficiency advantage over predecessors thanks to the addition of Intel® HT Technology. This technology lets each core run two threads, giving a quad-core Intel Core i7 the ability to run eight threads simultaneously (for example).

 

So how do the 32nm process and Intel HT Technology translate to real-world performance?  As one data point, Intel says that the new Intel Core i5 processors are twice as fast as the Intel Core 2 Duo parts they replace.  The power numbers also look good.  Initial benchmarks show that for a given TDP, the new Intel Core processors are about 20\% faster than equivalent Intel Core 2 processors.

 

The new processors also boost performance and lower power through the new Intel® Turbo Boost Technology.  Intel Turbo Boost Technology takes advantage of the fact that a processor’s power rating—that is, its thermal design power (TDP)—is based on a scenario where every Intel Core runs an intense workload.  This scenario rarely occurs in practice, so there is usually headroom between the TDP and instantaneous power consumption.  Intel Turbo Boost Technology senses this headroom and cranks up the clock speed when the processor is less than fully loaded.  The boost varies depending on the workload and the processor, but it’s common to see overall performance gains of 15\% or more.

 

For an even bigger burst of speed, Intel Turbo Boost Technology can temporarily cut the power to any idle Intel Cores.  This feature can produce impressive bursts of speed.  On the Intel® Core™ i7-620UE, for example, running Intel Turbo Boost Technology in single-core mode fully doubles the clock speed.  This feature lets you choose between multi-core operation at a “normal” clock speed or single-core operation at a boosted clock speed.  (See Intel’s YouTube video for details.)  This is a valuable choice for applications that are difficult to partition into multiple threads.  For these applications, you can write your code as a single thread and run it on a single sped-up core.

 

The following table lists the baseline and “turbo” clock speeds for each processor, as well as other vital specs:

 

Processor

Base clock speed

(GHz)

Turbo frequency

(GHz)

Cores /

threads

TDP

ECC

Intel Core i3-540

3.06

n/a

2/4

73W

Yes

Intel Core i5-520E

2.4

Up to 2.93 GHz

2/4

35W

Yes

Intel Core i5-520M

2.4

Up to 2.93 GHz

2/4

35W

No

Intel Core i5-660

3.33

Up to 3.60 GHz

2/4

73W

Yes

Intel Core i5-750

2.66

Up to 3.20 GHz

4/4†

95W

No

Intel Core i7-610E

2.53

Up to 3.20 GHz

2/4

35W

Yes

Intel Core i7-620LE

2

Up to 2.80 GHz

2/4

25W

Yes

Intel Core i7-620M

2.66

Up to 3.33 GHz

2/4

35W

No

Intel Core i7-620UE

1.06

Up to 2.13 GHz

2/4

18W

Yes

Intel Core i7-860

2.8

Up to 3.46 GHz

4/8

95W

No

†Intel HT Technology is not supported on the Intel Core i5-750.

 

The new processors also let you balance power between the CPU and the integrated graphics.  If you are running a graphics-intensive workload, you can boost the GPU clock while throttling the CPU.  Conversely, you can boost the CPU clock and throttle the GPU for workloads with limited graphics requirements.  This flexibility is quite handy.  Some embedded applications like digital signage are graphics-heavy but CPU-light, while others like telecom are CPU-heavy but graphics-light.  (Note that the Intel® Core™ i5-660 does not support this graphics clock scaling feature.)

 

The Intel Core i5 and Intel Core i7 also come with new AES-NI instructions that boost encrypted drive performance considerably. This feature is useful for applications where drive security is a concern, including retail kiosks and other unattended equipment.

 

Other notable features include the availability of ECC support on most of the new parts.  This feature is a must-have for military applications and a big plus for applications that require high reliability, such as telecom applications.

 

In short, it’s clear that the new Intel Core will be a popular choice for embedded systems.  Intel expects 200 design wins with the part, and a large number of modules, boards, etc. are available today.  The following table summarizes some example offerings.

 

Vendor

Form Factors

(and Product Names)

Target Markets

More

Information

Advantech

Mini-ITX (AIMB-270 and AIMB-280),

microATX (AIMB-580),

ATX (AIMB-780),

PICMG* (PCE-5125),

5.25-inch SBC (PCM-9593 ),

COM Express* (SOM-5788)

Telecom, factory

automation, military,

medical, gaming

bit.ly/8Y66la,

bit.ly/dwVYmh

Emerson

microATX (MATXM-CORE-411-B),

COM Express (COMX-CORE),

VMEbus (iVM7210)

Point-of-sale (POS),

industrial, medical,

military/aerospace

bit.ly/aUGNdf

Kontron

AMC (AM4020),

VPX (VX6060),

ETXexpress* (ETXexpress-AI)

Communications,

military/aerospace,

medical, industrial,

infotainment

bit.ly/9QocbO,

bit.ly/9fTENb

Radisys

COM Express (Procelerant CEQM57)

Medical imaging,

communications,

military/aerospace, test

and measurement

bit.ly/a83R48

 

All five companies mentioned in this blog are members of the Intel® Embedded Alliance.  Advantech, Emerson Network Power Embedded Computing, Kontron, and Radisys are Premier members; GE Intelligent Platforms is an Associate member. 

 

Kenton Williston
Roving Reporter (Intel Contractor)
Intel® Embedded Alliance
Editor-In-Chief
Embedded Innovator magazine