Intel has spoken of its near-future plans to tackle the mobile processor market, where it’s currently being soundly beaten by ARM, whose architecture is used on more than 90-percent of the world’s smartphones. While Intel has already had a go at grabbing some of the market for itself with the Medfield Atom chip, its limited adoption has seen it become largely forgotten.
So what’s next for Intel? Its plan is to forge ahead with the creation of much smaller and more efficient mobile systems-on-a-chip, this time built using a 22nm manufacturing process, which it currently uses for its Ivy Bridge PC processors amongst others. Intel’s current crop of mobile chips are built using 32nm technology, and by shrinking them down to 22nm, they would become the smallest of their type in the world.
All this was announced at the International Electron Devices Meeting which took place in San Francisco earlier this week. Reuters reports that the presentation used at the conference promises the 22nm Intel SoCs, “will be ready for high volume manufacturing in 2013.”
Drawing on experience
This makes it all sound very easy, and while Intel has a certain advantage over its competitors by already building 22nm chips – Qualcomm’s at 28nm and Nvidia’s at 40nm – it’s quite different to try and squeeze all the components not usually associated with PC processors, such as the memory and the radios, onto a single, tiny chip. Intel says its going to overcome this problem by employing the same 3D Tri-Gate transistors it uses in the production of its Ivy Bridge chips.
According to a report by Bit-Tech this would, “provide a significant boost in efficiency,” while an Intel spokesperson said 22nm Tri-Gate chips could offer hardware manufacturers with the option of increased performance from the same amount of power, or to keep the performance the same as 32nm chips while lowering power usage. As standby times and battery performance are becoming more important to us, this could be seen as a considerable benefit.
However, for all Intel’s noise, no other date except for 2013 was offered for when the next generation chips would arrive. Because it has taken nearly a year to launch a handful of phones using its Medfield chip, it’s hard to imagine anything much happening at a consumer level before 2014. Earlier this year, Intel Mobile VP Mike Bell said the Merrifield 22nm smartphone SoC would be out in 2013, saying, “This is a really big deal for us,” and that it would, “change the game for Intel in the smartphone world.”
It would be a Market Boost for Intel as Its planning to forge ahead with the creation of much smaller and more efficient mobile systems-on-a-chip,
ahmedkabir
The ARM marketing model is much different than Intel’s. ARM is an open architecture that can be licensed and mixed with other compatible building blocks to create custom SOC devices. So manufacturers can quickly design, simulate and fabricate exactly the architecture they want, Additional, the ARM cores are scalable and nobody is really even taking advantage of that aspect yet.
Intel, on the other hand, holds the Silicon design very closely and designs monolithic solutions because “Intel know what is best for you” … but they say it’s because they know how to manage the Silicon best.
The reality is we are in an environment where “Time To Market” is the chief competitive advantage. When product life spans are 18 months, you can’t wait 24 months for Intel to lay out the perfect piece of Silicon. SO even if ARM were an inefficient processor (NOT), a manufacturer can tweak a custom part, get into a market, make their money on the sweet side of the Bell Curve and move on to the next idea.
As for the smaller geometries, ARM can go there but there hasn’t been a need yet. ARM devices can be fabricated in all the best foundries (Samsung, TSMC, etc). Intel must go to 22 nm with their architecture just to compete with the current ARM efficiencies.
Sometimes I think Intel is just too big to play in the embedded processor market. They can’t “Bob and Weave”. The sad part was they had the ARM architecture once (StrongARM) and mothballed it because “they knew better than everyone else”. [Reference: Compaq iPaq PDA] It was almost like they bought the StrongARM to eliminate the competition.