Electronics Weekly Electronics Design & Components Tech News
But Samsung announced last week that it was churning out Apple processors on its 14nm line in Texas.
Hence Gassee’s question. He sees three possible answers to it:
“One is that the 14 nanometer process is woefully late. Deliveries of some Broadwell chips (the nickname of the next round of x-86 processors) are now slated for early- to mid-2105. Apple might feel that Intel’s process needs to mature before it can deliver 300M units.”
“The second is that Intel’s claim of a three-year technology lead might be less than reliable. Samsung could be closer to delivering 14nm chips than Intel would like us (and itself) to believe.”
“Or perhaps Intel sees Apple as a real adversary that’s intent on designing all of its own processors, even for laptops and desktops that are currently powered by x-86 chips. But even so, why not become the preferred fabricator?”
For Intel, the really worrying thing about this is if AMD gets to a 14nm process in a timely fashion. GloFo is said to be licensing the process off Samsung and, if GloFo can get it into production soon, then Intel could start seeing some unwelcome, and unusual, competition in the x86 market.
The tablet market is going the way of the 10inch notebook so it won’t be replacing anything, However the thought of a phone with a docking station is interesting. The problem I see is why carry all the hardware and performance needed to drive a 4k or 5k monitor ? Same for audio of course – who needs a 7.1 phone ? Instead put the performance needed in the docking station and keep the phone processor optimised for what it’s good at.
In any case this all falls down on the software. I don’t know why you think there is convergence but having just spent 3 weeks trying and failing to port a ,NET application to a Linux on Intel box I think there’s a huge way to go yet. Also my PC has around 200 applications installed and the hard disk image can be traced back over a decade, albeit with regular processor updates. But phones are replaced 2-yearly and I think people would baulk at re-installing all their applications, even if they are still available.
Very wise words indeed Stooriefit and even Intel seem to have caught up with that thinking. The “14nm” node is the first at which Intel has put a mobile processor at the front of the node – the Broadwell-M mobile processor is the first Intel mobile processor to be made on a leading-edge process.
My phone use is mostly as a data terminal, and my Galaxy S4 seems to perform data functions as fast as my iPad Air 1, so I have no beef with processor performance, Mike, but I’ll never say I don’t want more
We are less than five years away from the tablet or even phone replacing the office laptop.
You will have a phone or tablet powerful enough that a docking station on your desk gives you keyboard, video and mouse, wired network if needed, and access to the usual desktop applications.
The only real hurdle is getting enough convergence between desktop and mobile OS’s and that is happening in both Windows and OSX/iOS with every release. Office applications are becoming much more network/subscription driven now, so that fits in with this general direction of travel, and with universal binaries or .net framework processor architecture barriers are dropping too.
The user experience would leave a lot to be desired at the moment because we would miss all the background stuff we have got used to our desktops and laptops doing that our tablets and phones don’t have to worry about.
These top notch processors are needed so that the user experience of people’s phones doesn’t suffer too much as the convergence between mobile and desktop OS (and hence increasing workload on the mobile processor) happens. A tipping point on processor power and mobile OS capability is coming and then suddenly docking mobiles will be the norm.
People will only need PCs for really heavy lifting – coding, gaming, image and video processing, analysis and industrial computing etc. Code compilation, image and video processing and analysis could be farmed to a server (as is the case in even small operations now) which could be private or public, so the PC market will really shrink very soon.
This is one of the reasons your iPhone 3G runs like a dog with an up-to-date OS.
Can anyone show me a phone whose performance is limited by the processor ? Unfortunately nowadays it’s all about how many cores and GHz you run at, without any justification. Pure fairyland as David says.
Some hope! BPg, but one thing we will see is the performance of the ICs made on these three processes which will give us a clue on the relative merits of the processes. Costs, I suppose, we will never know.
Would appreciate to see the yield ramping as a part of the cost-margin equation in benchmarking among intel, Samsung and Tsmc (16nm) to validate the fairly-land argument.
Well if Samsung is giving mobile chips to Apple at cost, and we know Intel is paying people $4 billion a year to use its mobile chips, and we know the number on the node means nothing any more, and you say mobile chips don’t need 14nm but both Samsung and Intel are using so-called 14nm processes to make mobile chips, then we’re not just living in a marketing-led world, Mike, we’re living in fairy-land.
How about “or Samsung gave Apple the chips almost at cost such that even TSMC couldn’t compete”. Intel is miles ahead on technology but has never had the cost containment measures that Samsung and TSMC have honed over the years. At the end of the day no phone actually needs the performance 14nm can deliver – it’s all about marketing nowadays.
Very wise words indeed, Stooriefit. Intel, for a few quarters, showed how awful their mobile business is but now they’ve replaced the top guy and removed that transparency.
Another possibility is; given that we don’t really know what 14nm really refers to maybe what Intel are calling 14nm is much better than what Samsung are calling 14nm and commensurately harder to deliver?
Or maybe Apple aren’t prepared to pay Intel’s internal going rate /mm^2 for 14nm process. I know that Broadwell must have a similar effective end user price to the cost of an Apple processor, but you can bet that the Intel accountants have found a way to subsidise Broadwell-Y post-fab (using MFD etc) so that the fab overheads are covered. This protects their foundry margins with other customers but gets them into the market with their mobile technology which when taken in the round is certainly second best. Hiding this kind of shenanigans is why the mobile division was rolled up into a profitable part of the business after all.