Monthly Archives: January 2012

Geek speak

Intel vs. ARM in 2012

Since the beginning of the Technology Age, the introduction of portable devices as well as more advanced machines for solid working into the consumer market has seen a massive jump in the need for high-performance processors.

The market for processors for Internet connected mobile devices is evolving into a battle between the ARM and x86 architectures, with both architectures having advantages and disadvantages for particular applications. Both camps have been targeting the other markets; both are well entrenched.

The battleground is set in an era where historical mega trends of mobility and the Internet have morphed into the mobile Internet and Internet ubiquity. The integration of Internet access into mobile devices reaches well beyond simple email applications and web browsing. Mobile devices now must provide interactive communication through social networking and access to live, on-demand, and user generated entertainment. As a result, the mobile processors that power these devices must handle media processing, computing functionality, and an increasing number of wireless connectivity options into a usable form factor with all-day battery life. (CDRinfo The Hardware Authority, 2009)

Intel has long since been the leader in designing and manufacturing processors for business level machines where there is a high level of computing power required. The entire structure of its processors has been the core reason for its success as it has shown compatibility across the board even when it comes to different Operating Systems.

Over the past 25 years, Intel has risen to become the leading supplier of microprocessors for home and business computing, commanding a virtual monopoly in the market for desktop, laptop and server CPUs. (InfoWorld, 2009)

Intel has attempted a few times at mobile chip processing – Google TV was the last effort they had. It underperformed. They did launch the Medfield chip for Smartphone’s last year and it seems to have promise, as they have honestly put it all on one chip where previous designs didn’t.

 

Intel’s biggest success was the design and creation of the x86 architecture which surpassed all expectations.

 

Android is Intel’s main hope at gaining teeth in the market. The iOS and Windows Phone 7 are not going to want to go x86 until x86 trounces ARM as an option. The reason for this is, Android was designed from the get-go to be capable of being compatible with multiple forms of hardware. That is why it runs most applications in Java. It uses the JIT compiler to run Java applications as if they were native. So, in the case of Intel, they essentially just had to optimize the Java compiler for x86, and lo and behold, it ran 95% or more Android applications right off the jump. The iOS and Windows Phone 7 do not have this luxury, because both of those systems are built to specific hardware and have been optimized specifically for ARM, and both of their respective applications are developed with the intention of being on ARM chips. Android has a few applications that don’t run in Java, particularly games, but for the most part, Java is Android’s backbone, and will run fine on x86. Basically, if iOS and WP7 went all out on x86, they would tussle over developers and their application stores. Android essentially is the only major Operating System that can legitimately support both options.

 

Intel essentially has this Medfield chip, and it seems to perform as well as ARM chips clock speed to clock speed, if not faster. But that isn’t taking into account that ARM chips this year will be substantially improved. Overall, Intel won’t win over the mobile market with this generation. They will, at best, get their feet wet, where ARM won’t lose a whole lot this year. This year’s ARM chips are less power hungry, but also have LTE radios integrated on the chip, so will see HUGE power savings overall, of which Intel currently has no answer for.

 

So where will Intel benefit? Intel’s advantage is that they have a lot of capital, a ridiculously large amount at that. And their transistor technology trounces ARM. They are generally 3-4 years ahead of the competition. You will see that in Intel’s second generation chipsets, in which they will likely put out an offering that is both faster as well as more power efficient than comparable ARM chipsets. Where ARM this year will just be getting 28nanometer chipsets, Intel already has 22nanometer in their desktops (32nanometer in the mobile chipset for now). By next year, Intel will be down to 14nanometer. That sounds like small change, but you’re talking half, maybe a fourth of the power consumption at the same clock speed. So basically, x86 is the overall faster architecture, but more power hungry. BUT, their transistor technology will essentially let them muscle their way into having a fast, efficient chipset. I would expect those in 2013, which will be when it’s all out war for chipsets. 2012 is basically just the starting point.

 

However, people are quickly assuming that ARM Holdings is about to become the next Advanced Micro Devices (AMD), and like AMD they will simply fall off because they’re outmuscled by the moneybags, and I simply do not think that it is the case. Advanced Micro Devices is and was in a very unique position. x86 is effectively Intel’s brainchild, but because of old (like 1980s old) licensing agreements, AMD continues to make x86 chips. Advanced Micro Devices actually pioneered the 64 bit instruction set, which is currently why they are able to leverage maintaining such a license. AMD got the license to begin with because International Business Machines Corporation (IBM) required more than one supplier of x86 chipsets. So basically, Intel simply utilized AMD’s 64 bit instruction set, and mimicked the AMD design, and bested them with transistor technology that they simply could not compete with.

 

ARM Holdings is unique. They have generally the superior architecture in terms of power efficiency. The x86 architecture is a bit faster, but really ARM Holdings probably offers the best overall performance for the price (power). But in addition to this, ARM is also being helped by IBM, who supports the ARM architecture. So, the gap for who gets to compact, more efficient processors is getting a little bit smaller, as IBM’s research has very deep pockets as well. They are still a bit slower to technology than Intel, but nothing like what AMD suffers from.

 

Then we get down to the business models of both companies. Intel has, and will always continue to be a company that is closed off and controls their own designs greatly. They guard the instruction set quite well and including its proprietary. Intel does not license the x86 instruction set. You have to get it from Intel, or one of the old grandfathered in licensees like ARM or VIA Technologies.

 

ARM, however, licenses (for a fee of course) to anyone that would want to build a processor. This is why you see companies like Qualcomm, Nvidia, Samsung, and many others making ARM compatible chipsets. It allows for a lower cost to the hardware manufacturer (much cheaper to license the instruction set than to build one of your own) so you can hit lower targets in terms of overall cost. The other benefit is that it allows for variants, and hardware manufacturers can develop customized solutions for hardware. For example, putting more components on a System chip design, like Long Term Evolution (LTE) radios, the new 3-D transistors. It lets them cut costs and increase efficiency drastically. ARM chips are everywhere that you wouldn’t think, and have tons of custom solutions.

 

To further elaborate on the previous statement, the ARM ecosystem has unique and almost unrepeatable strengths – its universality and amount of partners: Qualcomm, Samsung, Texas Instruments, Nvidia – and that’s just the key ones that Microsoft has on board. There are stacks more. Even Apple is invested in this platform. Apple is a unique case as it now has its own chip design firm – it bought PA Semi who designed the ARM-based Apple A4 chip. Barring a simply earth-shattering seismic shift, Apple won’t be using Atom inside the iPhone 10 or iPad 5. ARM’s other benefit is that, while it designs the architecture, it uses a licensable model. That means that manufacturers, such as Nvidia with the Tegra or Qualcomm with the Snapdragon, can make their own design adjustments before manufacture. Snapdragon, for example, features Qualcomm proprietary GPU technology. They can also pick the best ARM processor to go in their products and that freedom of design and manufacture is something Intel won’t offer. (techradar.com – Deep into technology, 2011)

So essentially, you see Intel giving away processors at first, if only out of desperation to get teeth in the market. Predictions do not see them being cost-competitive with ARM off the jump. We also do not see them licensing out their architecture. We believe that their plan is the same as it was for desktops, to muscle their way in by throwing money at the manufacturing process and making superior chips by brute force. But, that plan could also backfire on them. Unless they make a clearly, undeniably superior solution, the manufacturers are not likely to want to pay Intel their premium to use their processors.

 

This leads us to the server market. Basically, the server market is 100% tied to Intel, dominated by them in the same way that ARM dominates the mobile chip market. So ARM finds itself in a similar situation that Intel is in with mobile, only inverse. Intel has these behemoth Xeon chips that offer mind-blowing performance. They have everyone on lockdown.

 

IBM and HP and Dell believe in the ARM architecture. They are working to nudge themselves into the server market. With the same essential goal of mobile; power efficiency. While that does not sound like a goal of the types of companies needing servers, just a few years ago, Microsoft was literally begging Intel to come out with servers with multiple Atom processors to try to cut their power costs. As everything goes to the cloud, more-so than native, power efficiency becomes crucial. Companies like Google, Amazon, Apple and Microsoft, are all going server heavy and offering cloud services, and that market is wide open. As more people adopt it, efficiency becomes imperative. Hence, we see ARM building a niche there. ARM has the architecture to utilize that. But in the total opposite of Intel, they need the hardware capability to make smaller, faster chips, where Intel is struggling to find a good balance between power efficiency and speed.

 

Much comparison has been made between the ARM and x86 Architectures since their release. For a trained eye, that of an Information Technology specialist, the list could go on but for the sake of those who are not savvy in the lingo of the essential geek, here is a simplified breakdown of the two products:

–       The advantages of the ARM architecture are design flexibility, graphics/multimedia Internet Protocol (IP), and processor footprint/level of integration.

–        The x86 architecture shines in terms of software compatibility and computational performance, which can sometimes be used in place of dedicated IP.

In a nutshell, ARM is a much smaller company than Intel, with a market capitalisation, at £8.4 billion, round about a tenth of that of the American giant. Whereas Intel has 90,000 employees, the Cambridge-based outfit has a mere 1,700. But many people see this company as the future of the semiconductor industry. Whereas Intel focuses largely on chips for computers, ARM’s strength is in portable devices such as Smartphone’s and tablets.

 

The reason for this is that ARM’s reduced instruction-set technology allows its chips to consume less power, making them ideal for portable devices. And the growth area in hardware at the moment happens to be portable devices.

ARM dominates the Smartphone market, with a 95% market share. But what is really exciting investors is that the business is branching out into other, lucrative areas. (The Motley Fool, 2011)

 

You’ve basically got two companies that previously were in niche markets, beginning to get to the point that they finally face off (again). And really, both have everything to gain and everything to lose. Technology and business pundits would vouch for either one but according to the recent trend in mobile chip technology, ARM seems to be the hot favourite. Intel is still the grandfather of microprocessors and financially speaking, it has nothing to worry about as it is the clear leader in its field. Now that the race for mobile chips is on, ARM is not slowing on its marketing rampage but is clearly eyeing Intel as they head towards the finish line.

 

In a blog published in 2010 comparing the ARM processors and Intel’s x86 architecture (more specifically, the Atom), a performance benchmark test was carried out and the findings are noted below: “The ARM Cortex-A8 achieves surprisingly competitive performance across many integer-based benchmarks while consuming power at levels far below the most energy miserly x86 CPU, the Intel Atom. In fact, the ARM Cortex-A8 matched or even beat the Intel Atom N450 across a significant number of our integer-based tests, especially when compensating for the Atom’s 25 percent clock speed advantage.

 

However, the ARM Cortex-A8 sample that we tested in the form of the Freescale i.MX515 lived in an ecosystem that was not competitive with the x86 rivals in this comparison. The video subsystem is very limited. Memory support is a very slow 32-bit, DDR2-200MHz.

 

Languishing across all of the JavaScript benchmarks, the ARM Cortex-A8 was only one-third to one-half as fast as the x86 competition. However, this might partially be a result of the very slow memory subsystem burdening the ARM core.

 

More troubling is the unacceptably poor double-precision floating-point throughput of the ARM Cortex-A8. While floating-point performance isn’t important to all tasks and is certainly not as important as integer performance, it cannot be ignored if ARM wants its products to successfully migrate upwards into traditional x86-dominated market spaces.” (Blog O’Matty, 2010)

 

Forget Intel versus AMD–that was a chip-maker battle of yesteryear, played out inside your desktop PC. Now the real CPU war is happening inside Smartphone’s and servers, where Intel is playing a desperate game of catch-up to ARM and a few new pretenders, and there’re billions of dollars at stake. (Fast Company, 2010)

Advertisements