backtop


Print 11 comment(s) - last by 91TTZ.. on Mar 1 at 1:53 PM

Partnerships with ZTE and LAVA means extra Atom-based handsets in 2012; but is it too little too late?

At the 2012 Mobile World Congress (MWC), Intel Corp. (INTC) talked about its increasing focus on the mobile market.

I. Getting in the Game

Intel has watched grimly as sales of mobile devices have raced towards sales of traditional personal computers and -- in the case of tablets -- potentially encroached on them.  Unlike ARM Holdings plc (LON:ARM) and its consortium of ARM chipmakers, Intel doesn't have a veteran smartphone or tablet chip line.  Intel has for some time tried to push out a smartphone chip, but has struggled to produce an x86-based chip with acceptable power performance.

But the company has fielded its first smartphone silicon, with the Medfield Z2xxx series of Atom CPUs.  Earlier this year the 2012 Consumer Electronics Show Intel demoed designs based on its first Medfield chip, the Atom Z2460 (single-core).  Intel also announced a long-term partnership with Google Inc.'s (GOOGpending acquisition Motorola Mobility.

The new smartphone system-on-a-chip features GPU and video encoding/decoding intellectual property cores from Imagination Technologies plc (LON:IMG), a mobile graphics chipmaker, who also works with many ARM chipmakers.
 
Intel Medfield
Medfield is the end result of a long process for Intel. [Image Source: Mobit]

The feel at the 2012 CES was that Intel was targeting a softer launch in 2012.  That feeling continued at the 2012 MWC.

Intel announced that samples of higher-clocked Atom Z2460s, with speeds up to 2.0 GHz would be shipping later this year, along with Atom Z2580, a dual-core design, and the 1.0 GHz Atom Z2000, a budget design.

The trio of CPUs won't hit the market in actual products until 2013.  These chips will be manufactured on Intel's mature 32 nm process, which deployed on the personal computer side in January 2011.

II. Die Shrinks

Looking ahead, Intel will be sampling 22 nm parts in 2013, but the language used indicates that they may not ship in actual product until 2014.  In a press release Intel writes:

Building on these 32nm announcements, Otellini discussed how the Atom™ processor will outpace Moore’s Law and announced that Intel will ship 22nm SoCs for carrier certification next year, and is already in development on 14nm SoC technology.
 
Where Intel gets itself in a bit of trouble is ambiguity.  Previous roadmaps showed 22 nm chips in 2013, so tech enthusiasts (and analysts) were likely expecting them.  Now the picture is a bit clearer, and it seems evident that while 22 nm samples are definitely in store for 2013, 22 nm products may not be.

Still Intel's process prowess is impressive.  It will soon become the first major chipmaker to place CPU designs the 20-22 nm node, with its 22 nm Ivy Bridge processor, which will reportedly launch in June.  If Intel can parlay its process knowledge and power-saving nanotechnologies -- such as 3D-FinFET transistors -- into smartphone designs, it could have a winner on its hands, despite its very tardy start.

On the other hand, the longer Intel waits, the more likely the ARM chipmakers are to capitalize, deploying more refined architectures and working on die-shrinks of their own.  Qualcomm Inc. (QCOM) and Samsung Electronics Comp., Ltd. (KS:005930) are both on the 28 nm node [1][2].  While it may take them a couple years to hit the next node, they will likely have until 2014 to field 20-22 nm products, if they wish to keep abreast of Intel in die shrinks.

III. Partnerships With ZTE, Lava; New HSPA+ Chipsets

Intel's partnership with Motorola will help to push Atom-powered Android phones, but Motorola is currently relatively weak in terms of smartphone sales.  On the other hand, Intel did announce what could be a crucial partnership with ZTE Corp. (SHE:000063), a Chinese phonemaker that is among the fastest growing phonemakers in the world.  ZTE sells a mixture of Android smartphones and traditional feature phones.

Intel also announced another key developing world partnership, pairing with India's LAVA International, Ltd.  Another fast-growing party, LAVA aims to ride the emerging wave of demand for smartphones in India, which is expected to reach 19.1 million units this year.

LAVA and Intel announced a new phone, the XOLO X900.  The phone will feature a single-core 1.6 GHz Atom Z2460 chip, a 4.03-inch multi-touch screen, an 8 MP camera, and support for HSPA+ and "all" 2G/3G network technologies.

Xolo X900
The XOLO X900 packs a 1.6 GHz Intel Atom Z2460 onboard. [Image Source: Intel]

The chip will use the XMM 6220 HSPA+ chipset.  The mixed chip will pack an ARM11-based baseband processor (courtesy of Intel's recent acquisition Infineon) built on a 40 nm process and a 65 nm digital RF transceiver.

Intel (and likely Motorola) will later this year launch products based on the XMM 7160 LTE (D: 100 Mbps; U: 50 Mbps)/HSPA+ (D: 42 Mbps; U: 11.5 Mbps)/3G/2G modem, another design from the Infineon unit.  The XMM 7160 is currently being sampled.  Intel also announced the XMM 7360 modem, which will be similar to the XMM 7160, but lack LTE support.

IV. The Roadmap

Summarizing, this year Intel will ship or sample a variety of mobile chipsets, as well as its Medfield Z2460 CPU.  Motorola, ZTE, the Lenovo Group, Ltd. (HKG:0992), and LAVA will all ship Atom-based Android smartphones in 2012, but don't expect high volume or multiple designs.

In 2013 the lineup will expand slightly, with the single core 1.0 GHz Atom Z2000 low-end CPU, faster clocked Atom Z2460s, and the dual-core Atom Z2580.  Production will continue to be 32 nm.

Finally, in 2014 (or late 2013) expect Intel to finally go full blast with 22 nm technology.  It's a long road ahead, but Intel is adamant that it will see mobile success.

We won't know the company's fate until thorough benchmarks emerge later this year, and even then the true measure of success or failure will be in sales volume.  Intel's approach of targeting developing markets is either clever -- or desperate (or perhaps a bit of both).  Which it happens to be should clarify as the year goes on.

Early figures make Intel look power competitive with 40 nm ARM Cortex A9 cores, but the problem is that Intel faces a moving target -- by the time it ships product in Q2/Q3 2012, more powerful, power-efficient 28 nm ARM Cortex A15 will be upon the market.  And experts like AnandTech's founder Anand Lal Shimpi believe that Cortex A15 should be able to beat Medfield in performance, if not power.

Sources: Intel [MWC press release], [XOLO X900 press release], [XMM 6220 press release], Anandtech



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

It's never too late
By tayb on 2/27/2012 5:08:51 PM , Rating: 5
With hardware it's never too late, this isn't software. If Intel releases a competitive chip at a competitive price they will find partners regardless of when that happens. With the amount of revenue Intel is generating I certainly wouldn't bet against them.




RE: It's never too late
By retrospooty on 2/27/2012 5:17:37 PM , Rating: 2
Yup, throw in another process shrink to 22nm and maybe a bit more work on power efficiency and they are right in it.


RE: It's never too late
By someguy123 on 2/27/2012 5:44:56 PM , Rating: 2
They definitely have competitive specs on paper moving to 22nm and finfet, but the delay is much longer than I had expected. I can understand the delays considering they're the only ones with a seemingly working finfet process, but it's a pretty big window for ARM to come up with something competitive.

Then again, looking at ARM's current trend of stacking cores while gradually increasing power draw with every flagship, intel may have a bit of breathing room here.


RE: It's never too late
By fteoath64 on 2/28/2012 4:43:54 AM , Rating: 2
Nope. Its too late for Intel. The architecture is wrong for mobile and its scaling does not make sense either. No matter how much money they throw at it, this will not be competitive because ARM A15 is already here!.
The Snapdragon S4 shows remarkable performance for similar power consumption of the previous generation architecture. It scales to 2.5Ghz and is on 28nm node going smaller. The gpu also leads another step ahead, so Atom is losing big time here.

The volumes of A15 and Krait will speak market share for itself this year in smartphones and tablets. Intel should get an ARM license and innovate from there. There is still time unless they wanted to wait 2 more years of Atom suffering to "get it".


RE: It's never too late
By 91TTZ on 2/28/2012 1:05:52 PM , Rating: 2
It's not too late for Intel.

Right now PC and server processors are Intel's bread and butter so the mobile space is just a hobby for them. It shouldn't be surprising that they aren't putting their full weight behind the effort. If they really needed to they could turn the heat up on ARM and force them out of the market, but they probably don't want any antitrust action against them. Intel has such engineering and manufacturing capacity that they could change direction and introduce a new architecture if they wanted to.

You need to keep in mind just how much larger and more powerful Intel is than ARM. Intel has about 80x more revenue than ARM and employs 60x as many people. They could eat ARM's lunch whenever they wanted to.


RE: It's never too late
By someguy123 on 2/28/2012 5:12:51 PM , Rating: 2
how could it possibly be too late when they aren't even riding on that market? their desktop/server market are doing just fine.

All they really need to do is get the lower node processes working and they'll have a comparable processor. x86 bloat becomes less and less of an issue as you scale down, while maintaining legacy support. if they can't produce in time to compete, nothing will change and the market will continue to use ARM's risc designs.


Pedal to the metal
By Hector2 on 2/27/2012 6:51:58 PM , Rating: 2
Obviously, Intel has pulled out all the stops to catch up. After being dismissed by ARM and the analysts, they came out with a shocker last month when, in one huge leap, they just about wiped out the power & performance advantage that ARM has had until now and they did it with a single Core. While ARM was caught off guard, it's apparent now that this has been Intel's plan all along.

I think Intel has bigger cards yet to play in their hand than does their competition, and I'm betting that their next giant leap with more core and 22nm will leave ARM behind trying to catch up. I doubt that ARMs 4-core will be enough.

Amazing and fun to watch unfold.




RE: Pedal to the metal
By 91TTZ on 2/28/2012 1:15:02 PM , Rating: 2
quote:
After being dismissed by ARM and the analysts, they came out with a shocker last month when, in one huge leap, they just about wiped out the power & performance advantage that ARM has had until now and they did it with a single Core. While ARM was caught off guard, it's apparent now that this has been Intel's plan all along.


Yup.

This is similar to what happened when AMD captured the speed crown from Intel: Intel devoted more engineers to the cause, implemented that tick-tock strategy and quickly overwhelmed AMD. In hindsight it wasn't really ever a race, as Intel is so big that they could have crushed AMD when they wanted to. But Intel's biggest opponent isn't AMD or ARM, it's antitrust legislation. They want to manage their success wisely without killing their opponents, otherwise the government will attempt to break them up like they tried to do to Microsoft.

ARM has 1,700 employees and Intel has 100,000+. Did anyone really think that Intel would have a problem designing a competitive chip?


RE: Pedal to the metal
By ilt24 on 2/28/2012 4:32:07 PM , Rating: 2
quote:
ARM has 1,700 employees and Intel has 100,000+.


ARM has 1700 employees working on creating a set of portable IP. Then there are many dozen companies who license the IP that create various processor designs. These designs are then taken to a foundry to be produced...so Intel is competeing with a lot more than 1700 people in the phone/tablet space.


RE: Pedal to the metal
By 91TTZ on 3/1/2012 1:53:44 PM , Rating: 2
As far as I understand it, the actual core of the CPU is designed by ARM. If Intel wanted to catch up to ARM, they could throw a hell of a lot more than 1,700 engineers at the task to come up with a superior design.


Can't wait
By Church of Dirac on 2/28/2012 2:03:20 AM , Rating: 2
Finally a step towards that mythical "convergence device". I enjoy my Motorola Atrix webtop even with its shortcomings. A phone (or tablet!) running real x86 Windows 7 could change the game and knock out ipad




"If you mod me down, I will become more insightful than you can possibly imagine." -- Slashdot














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki