What a bunch of drivvel
I expect the odd moronic commentard on ElReg, but this thread is mostly crap. Clearly what is needed is a comment or two from someone who actually develops drivers and has worked in the field.
On openness...
ARM doesn't make chips. They make cores. They do **huge** open source work for helping Linux and gcc etc work better with their cores. For example, ARM are a major player in http://www.linaro.org/. They don't even mind that the Linaro compiler improvements are even helping improve the compilers for x86.
ARM does not, in general, implement drivers. Why? Coz they don't make the SoCs (a mix of cores + peripherals) and the drivers, for the most part, are for peripherals. It is thus up to the peripheral/chip makers to release drivers. Most (eg. TI and Freescale) do an excellent job. Some (eg. the people doing the chip in Rasberrypi) not so good. But none of this has ANYTHING to do with ARM.
Having implemented a debugging engine for ARM, from ARM specs, I can assure you that ARM does a damn good job of detailing how their cores work. They need this documentation because their customers need it to integrate ARM-based chips.
Intel, OTOH, does not sell cores. They sell whole SoCs. They thus need to produce all the drivers for the SoC or nobody can use the SoC features.
And therein lies part of the problem for Intel. Since Intel builds the whole SoC, they can really only pitch their offering at a few markets because different markets need different peripherals. ARM does not make chips, but tens of vendors make hundreds of different variants of ARM-based chips.
You think that Intel-inside PC is Intel based do you? Well think again - it has many more ARM cores in it than Intel cores. Just a hard disk driver will typically have 2 or 3 ARM cores in it. Then there's the wifi/bluetooth module and a swag of other chips.
Probably the major problem for Intel is that their business model is based on high margin chips: spend hundreds of millions on chip development then take a tablespoon of sand, cook it and make a chip you sell for $100 or so, then sell a few million. The ARM-based business model is different. The ARM chip vendors spend maybe $10M building a new chip then sell tens of millions of them for $20 (or even down to 50c for the low-end ARM parts). That makes it almost impossible for Intel to manoever in the same market.
Worse still for Intel, the place that these new Atoms might actually work is in tablets and laptops - displacing higher price (and higher margin) Intel parts. That means they're really cutting their own throats by replacing a $100 chip with a $60 chip. Still, I suppose, that is better than giving up and letting the ARM parts take over completely which would give them $0.
Thus it is really hard to see that Intel is going to make any inroads. At best they're fighting a rear guard action which will keep them in business for a bit longer.
Biting the hand that feeds IT
