Being the nerd I am I’ve kept a keen eye on all the
happenings at IDF 2011. The usual things have held my interest – Intels Ivybridge
‘tock’ architecture and X79 ‘Sandybrige E’ in particular. However, the biggest
surprise of the show was the announcement in Intel CEO Paul Ottelinis key note
that starting next year Android will be optimized for x86. For those of you
unfamiliar, X86 is the instruction set that computers in the traditional sense
of the word have used since around 1978 (though now in a highly modified form).
Intel has long since made clear its intentions to break into the mobile space. At this point, it’s worth running over why. Near domination of the PC CPU market, combined with slower than usual sales of laptops and PCs mean that new revenue streams must be sought. In turn, the reason for this reduction is a decreased need for ‘moar powa’ and a cannibalisation of the market by at first netbooks, and now tablets.
Intel has long since made clear its intentions to break into the mobile space. At this point, it’s worth running over why. Near domination of the PC CPU market, combined with slower than usual sales of laptops and PCs mean that new revenue streams must be sought. In turn, the reason for this reduction is a decreased need for ‘moar powa’ and a cannibalisation of the market by at first netbooks, and now tablets.
However, x86 was never designed with power efficiency in
mind, and this is why ARM has become so very dominant in the mobile computing
space. Intels newest ‘Sandybridge’ architecture is very power efficient in
terms of ‘performance per watt’ but still remain wholly unsuitable for tablets
or mobile phones requiring heavy, all day use.
So for x86 people to be banding about not just the word ‘compatible’ but
‘optimized’ in regards to Android is surprising.
I had always assumed that Intels inevitable transition to mobile would be with Microsoft or Meego Linux as its bedfellow rather than Google. The Atom lines newest iteration ‘medfield’ does have the advantage of 32nm lithography over its Arm rivals 40nm chips, and it is perhaps this that negates the disadvantages of using a less efficient instruction set. In the long run, Intel does appear to be set on sticking to x86 in the mobile space. To do this it will either have to spend on fabs to keep its lithographic process ahead of its rivals or it’ll have to develop complex additions to the instruction set to compete with ARM. While competition in any market is good, should Intel and x86 succeed in the mobile space, their unflinching domination of desktop CPUs leaves me (perhaps prematurely) concerned for ARMs ability to compete against the chip juggernaught.
I had always assumed that Intels inevitable transition to mobile would be with Microsoft or Meego Linux as its bedfellow rather than Google. The Atom lines newest iteration ‘medfield’ does have the advantage of 32nm lithography over its Arm rivals 40nm chips, and it is perhaps this that negates the disadvantages of using a less efficient instruction set. In the long run, Intel does appear to be set on sticking to x86 in the mobile space. To do this it will either have to spend on fabs to keep its lithographic process ahead of its rivals or it’ll have to develop complex additions to the instruction set to compete with ARM. While competition in any market is good, should Intel and x86 succeed in the mobile space, their unflinching domination of desktop CPUs leaves me (perhaps prematurely) concerned for ARMs ability to compete against the chip juggernaught.