Intel is scaling down its chips, while Qualcomm/ARM is scaling up.
If tomorrow Qualcomm (or somebody) decided to roll out a 200w TDP 4.0GHz chip. Likely the best they could do is throw a lot of cores at the problem (enter Oracle's T-5 SPARC). Which failed, the market wants parallelism, but they still demand a certain threshold of single threaded processing speed.
Qualcomm/ARM has to fight dragons like branch prediction, caching, multi-issue out of order pipe-lining, etc. if they want to compete against Intel. And these are Intel's crown jewels, they aren't going to give them away without a fight.
While when Intel wants to compete with Qualcomm/ARM they just shoot their processor power budget in the knee caps and start ripping components out (figuratively, not literally).
Qualcomm doesn't need to scale up. Qualcomm is slowly scaling up by releasing ever more powerful Snapdragons. Eventually, Qualcomm's chips will be fast enough to do everything we could reasonably want (or at least most things that most people could reasonably want). It could almost be argued that the majority of people (minus tech nerds and a few other people with special use cases) can get by just fine with a smartphone plus a tablet.
doesn't make any sense to me. mobile devices are like the new PC industry all over again..they become more and more powerful with each generation. only now there's a power budget handicap unlike on PCs. in time to come, they will probably enter intel territory in terms of power budget and compute. the problem then will be, has the world switched to ARM software, or are we still using x86-64? its a cop out, but i imagine the world is not likely going to replace essentially all its software anytime soon.. i mean, there could be ports, but as long as i have a x86-64 machine and the software works, why would i switch? unless the ARM port is much better? i fail to see how its going to happen.. all in all, intel just has to wait for mobile devices to catch up with existing PCs..if i were them i'd fund all sorts of battery tech to help mobile computing on its way...
separately, i thought intel always was ARM capable only they made hardly any ARM chips(i think they sold strongARM a long time ago)..and also a big difference between the ARM ecosystem and Intel is Intel is a fab as well..and a darn good one.
the only thing(ok this can't really be true) Intel doesn't seem to be able to do is make a good graphics processor..
> It’s actually only a matter of time before Apple releases a laptop based on their A-series of processors (possibly as soon as next year, but definitely by 2017), and from that point, the rest of of the PC market will slowly follow. In the end, Intel will eventually be driven out of the consumer market
Is this assuming that all Windows software will be replaced by OS X, or that all Windows apps will be ported to ARM?
Windows apps will be ported to ARM. Microsoft already supports an ARM version of Windows, so this doesn't seem so far-fetched. Also, it is possible to build universal binaries supporting two architectures (Apple did this during the PPC-Intel transition).
Windows RT tablets didn't do well because they were too slow and too expensive. They were basically pitched as full computers, but the chips were too slow for that. The Surface Pro was only a bit more expensive, and offered much better performance. But performance of ARM is rapidly improving, to the point where it will soon be sufficient.
And Chromebooks have always been available in x86. It's easier for notebook manufacturers to convert low-end windows notebooks to Chrome OS than to design ARM-based systems from the ground up. From what I understand, ARM Chromebooks have sold pretty well.
Intel is scaling down its chips, while Qualcomm/ARM is scaling up.
If tomorrow Qualcomm (or somebody) decided to roll out a 200w TDP 4.0GHz chip. Likely the best they could do is throw a lot of cores at the problem (enter Oracle's T-5 SPARC). Which failed, the market wants parallelism, but they still demand a certain threshold of single threaded processing speed.
Qualcomm/ARM has to fight dragons like branch prediction, caching, multi-issue out of order pipe-lining, etc. if they want to compete against Intel. And these are Intel's crown jewels, they aren't going to give them away without a fight.
While when Intel wants to compete with Qualcomm/ARM they just shoot their processor power budget in the knee caps and start ripping components out (figuratively, not literally).