Chipzilla just won't quit: Intel touts 64-bit Atoms for Android phones, tabs Intel has launched one new 64-bit, Android-capable Atom processor platform – "Merrifield" – taken the wraps off a future one – "Moorefield" – and introduced a new LTE modem in its latest bid to crack the mobile market as well as position itself for the buzzword du jour, the Internet of Things. "Today we are announcing leading …
Well, no-one's gonna make a 64bit OS if there's no chip to run it. And vice versa - until now. Since there's probably no architectural problem to compile Android/FF/Tizen/Windoze/etc to run with this CPU - why not? There's probably more overhead with 64bitness as it is now but I'm expecting tablets to gain more processing power and memory and such that 64 bits will make sense.
Iphone 5s is 64-bit. Are there benefits to that? - I don't know. But are you actually suggesting that "640K 32 bits ought to be enough for anybody"
"I co-wrote the following article which got journalistified :-P a bit but the gist of it is there:"
Hadn't seen that, a nicer read than Charles Arthur's related article a few days earlier.
If you still have contacts at the grauniad, can you get someone to re-educate Charles re some of his embarrassingly hilarious comments re 64bitness in his article (and in his comments attempting to defend his misunderstanding when readers attempt to clarify matters, e.g.
http://discussion.theguardian.com/comment-permalink/26938570 and
http://discussion.theguardian.com/comment-permalink/26952007
"All modern OSes on Intel x86 use flat address spaces. Segmentation is flattened."
I think you might have missed my sarcasm. PAE is a form of segmentation. Of course it's preferable to have the CPU's native word size be the same size as the addresses, or bigger. There is minimal overhead to moving a CPU core to 64-bit. Back when AMD did it in 2003 (?) they claimed that it only added 10% overhead to their transistor count. Surely now that CPU cores take a relatively smaller amount of die size on a typical processor, the overhead is less than 10% and probably just a single gate delay or two. There's really no reason for anybody to be using a 32 bit CPU for any kind of non-microcontroller application.
What do you want to bet that Intel helped fund it, and helped select the tests behind the scenes?
They pull this every time they release a new chip, and then when someone finds the off-brand phone using their chip and tests it, they find it isn't close to the claims being made about it. I'm surprised anyone even bothers to review them anymore, after they've fail to live up to the cooked benchmarks time and again.
"and we're the only company currently shipping 64-bit processors supporting multiple operating systems today, and capable of supporting 64-bit Android when it is available."
Well, ignoring AMD and probably several other obvious examples, haven't there been 64 bit computers since the 70s (60s?) that were able to run multiple operating systems? Have I missed something?
"haven't there been 64 bit computers since the 70s (60s?) that were able to run multiple operating systems?"
In 1993 (August, from memory?) there was a special issue of Byte magazine focusing on 64bit computing.
On the front cover were Alpha, MIPS, PowerPC, and SPARC. Can't remember if IA64 was there.
All were soon able to run some kind of UNIX derivative. Some could also run a proprietary OS (eg VMS on Alpha).
Most were soon able to run Windows NT (till Redmond pulled the plug from the multi-vendor support in NT), though obviously back then it was a 32bit OS on a 64bit chip.
So you're not missing much, certainly you're closer to reality than Intel's spinners are.
"and capable of supporting 64-bit Android when it is available."
Are AMD shipping an AMD64 core that works with Android? I know that really Android should be portable to AMD's chips but do they have a 64 bit mobile chip with Linux support and Android driver support? If not then what Intel have said there is factually correct.
"Are AMD shipping an AMD64 core that works with Android? I know that really Android should be portable to AMD's chips but do they have a 64 bit mobile chip with Linux support and Android driver support? If not then what Intel have said there is factually correct."
Oh, did the Intel person say MOBILE chip and/or MOBILE operating system? I didn't see either specification mentioned.
Either way, AMD certainly makes the (64 bit) processors for some tablets that I'm sure can run both Windows 8 and Linux.
"Either way, AMD certainly makes the (64 bit) processors for some tablets that I'm sure can run both Windows 8 and Linux."
Neither of those is Android. Do AMD ship an Android support package for their chips? Linux support is a bit different to Android support as Android has it's own driver system for graphics, hw accelerated video.... It's possible that AMD do have something but Atom is the only officially supported x86 platform for Android as far as I can tell. So what the Intel person said is factually correct and you're just picking hairs/quote mining.
"Neither of those is Android. ... So what the Intel person said is factually correct and you're just picking hairs/quote mining."
I don't think it's being too pedantic if I don't feel like making the leap from "the only 64-bit processor that runs multiple operating systems" to "the only 'mobile' 64-bit processor that runs multiple 'mobile' operating systems, one of which must be Android."
You can make anything factually correct if you just change all the words around.
>I don't think it's being too pedantic
You're making it seem like they made a wild claim that was untrue. They didn't .. they can run multiple operating systems today and they should be able to make their support for the current x86 Android work for amd64/x86-64 Android when it comes around. The point being made is that they have a product that is somewhat future-proof not that that they are the only 64bit game in town.
>You can make anything factually correct if you just change all the words around.
You don't need to change the words around to make it factually correct.
" ... The point being made is that they have a product that is somewhat future-proof not that that they are the only 64bit game in town."
Uh, what quote are you reading? Because the Intel woman DID say that they were the _ONLY_ company shipping 64 bit processors that run multiple OSs.
"and capable of supporting 64-bit Android when it is available."
Ah, I see. You're fixated on the part of the quote that I wasn't talking about, never argued with, is factually correct, and I only included because it was at the end of the sentence. "Thanks" for wasting everybody's time.
Can you take a intel Atom silicon and integrate it with the graphics, mobile chipset, and all the other gubbins that make a mobile phone.
You can take a Arm processor core(s) add customised silicon round the outside and send it to a Fab.
This is the real reason that Intel has not made much of a dent in the mobile market not some made up bench mark
>You can take a Arm processor core(s) add customised
>silicon round the outside and send it to a Fab.
If you can do that for ARM why would you not be able to do it for any other architecture?
I know it's hip to love ARM and hate x86 but there isn't anything that's amazingly special about ARM that isn't available in or possible to do with at least a few of the hundreds of architectures that are out in the wild.
>Uh, because Intel's architecture is proprietary and they will not sell you the design
>so you can go off and fab your own chips with your own modifications
There are companies that aren't intel fabbing X86 based SoC's.. And yes, ARM is a fabless semiconductor company that licenses their design but there is nothing special about the ARM design that would mean that Intel can't build similar mobile SoCs based around the Atom core. You could argue that the ARM designs are significantly less complex than modern x86 chips and thus are easier to integrate into low power SoC designs but I would argue that's becoming less and less the case as ARM is trying to get to a point that isn't "laughably behind" x86 on performance.
Intel Atom Processor Z2420
…
Android 4.0
From here:
http://www.maplin.co.uk/p/prestigio-multiphone-5430-mobile-phone-white-a72lt
and all for 199.99 of your earthling pounds (better grab it (earthling pounds that be) just in case UK Treasury loses its much anticipated future oil revenues no?)
Here's hoping they got a serious commitment to Linux GPU open APIs from Imagination Technologies. That will be critical to their success with these chips, and IT has been reluctant on that score in the past. "It's not our fault the GPU vendor won't provide that info" is not going to work. Ever. You chose the GPU. They didn't sneak into your engineering lab and insert it without your knowledge. It is your fault if the thing has no decent software support.
In a recent "flier" https://ilms.intel.com/aprimoportal/OutboundMessage.aspx?A=87144bbd78aff09964cca5937de9609c&O=664c8a60aa01db85ba39dd5c572028aa&T=b333208886639fd5&D=194e13319b5d48d29f3b706573901c28&M=b333208886639fd5&S=8aac5c2993895f67bc793c6121b4df98&I=3e7925015c832b66870e6ac6b792d7f6&MSGID=8e76f14230b53d645529482df90c4e58 , Intel quotes a great feature that makes them suitable ("Perfect") for embedded,
That feature is segmentation
I am not making it up ...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
