Intel: Android 4.0 has x86 optimizations

Ice Cream Sandwich devices with Intel inside are on the way. Barely two months after teaming up with Google to tune Android for Atom CPUs, Intel has revealed that the latest Android release already includes x86 optimizations. IDG News has the scoop:

The OS is ready to work with tablets and smartphones based on Intel’s x86 mobile processors for tablets and smartphones, Intel said this week. The first smartphone based on an Intel chip will reach the market in the first half of next year, Intel has said. Intel moved away from its own MeeGo OS in favor of Android, and the chip maker has been working closely with Google on developing a version of Android for Intel-based smartphones and tablets.
"Ice Cream Sandwich includes OS optimization for x86, so Intel architecture-based devices can support it," said Suzy Greenberg, an Intel spokeswoman.

The first Intel-powered Android handsets should feature Medfield, a 32-nm Atom chip Intel has tailored for not just smartphones, but tablets, as well. At Computex in May, Intel said Medfield-powered tablets will be thinner than 0.35" and lighter than 1.5 lbs. That means they could be comparable to today’s 10-inch, ARM-powered offerings.

Medfield should be in production this year. According to a CNet News piece posted in June, though, the chip won’t actually be out until the first half of 2012. (Thanks to Liliputing for the tip.)

Comments closed
    • pogsnet1
    • 8 years ago

    Funny they compare both cpu with physical weight which is less relevant, what matters most is the TDP.

    • Redemmed
    • 8 years ago

    Here we go again the RISC vs x86 design wars, this should be more interesting

      • srg86
      • 8 years ago

      I’m bored of it personally.

    • bcronce
    • 8 years ago

    While I love Intel’s engineering(not their lawyers/marketing), I hope they don’t mess up this mobile platform. The original Atom CPU was competitive, but the platform(chipset/etc) was horrible for power. It’s not often you see the CPU being only 1/3 of the power draw.

    • Jigar
    • 8 years ago

    If this is happening, AMD will also rejoice.

      • dpaus
      • 8 years ago

      Unless, of course, Intel has, um, ‘optimized’ it to run on [i<]Intel[/i<] processors. As in, 'only'. Not that they'd ever do anything like that, of course. Also, I've got to wonder just how much work was involved; Android is based on the very same Linux core as WebOS was, and I remember various parties getting WebOS running on desktop PCs within a few weeks of it's release - and with far fewer resources available to them than Intel has. I haven't followed the Android camp to the same depth; does anyone know if a similar feat was acheived there already? EDIT: just to be clear, I meant that the hobbyists had WebOS [i<]booting[/i<] on desktop PCs, not just running as a process.

        • Zoomer
        • 8 years ago

        Set a different compile target, write drivers, poke around code to see if there’s anything to be changed?

        That said, why anyone would use x86 on android is beyond me. Seems like a lose-lose proposition, unfortunately. The only reason for x86 to exist is the large software install base; it is a horrible bloated mess otherwise.

          • srg86
          • 8 years ago

          Do you want ARM to have a monopoly in that space? I see this as good extra competition. Everyone hates on Intel and x86 when it comes to these, by if Intel can get similar performance per watt (that’s an if) then why not have extra options.

          Everyone seems to hate Intel having a monopoly, yet seem to want ARM to have one (through its licensees, at least there’s competition there). I don’t get that.

          Having Intel and ARM (through its licensees) battle it out in all sizes of computers can only be nothing but good for the market (assuming there’s no shenanigans, I’m just thinking of pure competition here). If AMD can stay in as well, all the better.

            • Corrado
            • 8 years ago

            I think ARM has enough competition with itself. TI, Samsung, Apple, nVidia, Qualcomm and some other smaller players are all competing.

            • srg86
            • 8 years ago

            They all need ARM licenses though, that’s where I see a monopoly for ARM.

            Of course anyone who wants to make an x86 has to at least have a cross licensing agreement with Intel, do the two compare?

            • Corrado
            • 8 years ago

            RIght right, I’m thinking more along the lines of performance competition and lighting a fire for R&D.

            • NeelyCam
            • 8 years ago

            Without non-ARM competition, ARM can hike the license fees on a whim. And they have.

            • raddude9
            • 8 years ago

            not again… They have never hiked their fees. They just introduced very slightly higher fees for their most advanced chip designs.

            • srg86
            • 8 years ago

            Without going into whether they have already or not, it’s still a possibility if there were no non-ARM competition, it may not have happened so far, but the potential is there.

            Competition keeps everyone honest.

            • NeelyCam
            • 8 years ago

            So, following your argument, it seems you’d be perfectly happy if Intel had a monopoly and decided to sell ivybridge chips at 50% higher price than sandybridge? You know, because they are new and most advanced and all that..?

            I find this double standard of ARM proponents just baffling…

            And regarding what the “very slightly higher fees” mean:

            Here’s a link that says ARM royalty has been around 1% traditionally, and now is going to about 1.1-1.2%.

            [url<]http://www.eetimes.com/electronics-products/processors/4114549/ARM-raises-royalty-rate-with-Cortex[/url<] That's from almost two years ago. Now, if you google "ARM q1 2011 results", and click on the "quick view", you see some slides mentioning that new chips yield 2% royalties. Goldman Sachs analysts also points out this sudden increase in royalty percentage: [url<]http://ftalphaville.ft.com/blog/2011/02/14/487191/a-tale-of-two-tech-companies/[/url<] I wouldn't call doubling your royalty percentage as a 'very slightly higher fees'. How can they do this? Simple - they are currently the only game in town, and they are ruthlessly abusing their monopoly position. X86 will hopefully break this monopoly, and make ARM think twice about hiking their royalties again. Competition is good, right?

            • raddude9
            • 8 years ago

            But following your argument. Intel will have a monopoly everywhere, which is far worse than having two competing monopolies (although different market segments).
            1.2% on the cost of a chip is still very low and the reason they are still being used is that the resultant chips are still vastly cheaper and more highly integrated than intels cheapest chip (currently $43 for a lame Atom).

            And a 50% higher price is not the same as a 50% higher royalty, I have a feeling you don’t understand this. Let me explain in simple terms. A 2% royalty on a 50$ chip turns it into a $51 chip, whereas at 1% royalty the chip becomes a $50.50 chip. Now if somebody were to increase the price of a $50 chip by 50% you would get a $75 dollar chip. Sure, the 2% royalty is a 100% increase on a 1% royalty, but it only looks big because the 1% was so small. If ARM had always charged 10% and they adopted an 11% royalty on their new chips, that would mean that they only increased their royalty by 10%, and I guess that would be fine with you because it’s nothing like the 100% figure that you are so obsessed with. So, even using your 2% royalty figure, which I have never seen before, and I question it’s legitimacy, it’s not a big increase, Only the mathematically misguided would think that it was.

            Anyway, ARM are not the only game in town, as you yourself mentioned, MIPS chips are making progress in the smartphone/android market, and it’s to the likes of them and PPC that you should look to for competition, both are making progress in the Android marketplace. Intels monopoly is deisgned to keep the cost of entry to the world of computing high, they don’t make cheap chips because they would not be able to afford their fancy fabs.

            • NeelyCam
            • 8 years ago

            Maybe you don’t understand that monopoly is monopoly. In low-power chips Intel doesn’t represent monopoly – instead, they represent competition. Somehow you consider [b<]Intel[/b<] as something [i<]bad[/i<] regardless if they are competing for extending their supposed monopoly or breaking someone else. Hence, double-standard. Normally, in a competitive situation, ARM would've been forced to cut their royalties from 1% to 0.9%, instead of being able to hike them from 1% to 2%. And you as a consume will pay the difference. If the status quo remains, that royalty will become 5% or beyond. Wake up. Competition is good. ARM has none right now. X86 is about to challenge ARM, and I really don't understand why you're unhappy about that.

            • raddude9
            • 8 years ago

            Why am I unhappy, same reason as you, because Competition is good, but I’m looking at the bigger picture. There is very little competition in the Desktop processor space, x86 pretty much has 100% of that since Apple moved from PPC. There is some competition for processors in the server space, but x86 takes the lions share with the last remaining RISC (PPC/SPARC) holdouts losing market share consistently. And because AMD is quite ineffectual x86 means Intel. Intel used to have none of the server space, it used the profits it gleaned from the desktop monopoly to muscle it’s way in. I don’t want that to see that happen again on the mobile CPU space.

            You want more competition in the mobile space? I don’t get it, are you happy with the utter lack of competition in the Desktop/Server space, a market that is far larger and much more lucerative. Wouldn’t it be much better for ARM to use their advantage in the mobile space to give us competition in the much larger desktop and server markets, rather than for Intel to put the squeeze on the only serious competition it has seen in a decade in a market segment that is far smaller than the segments it currently dominates.

            So, I see you as having the double standards, you want competition in the mobiles CPU space but are happy to go on living with your Intel monopoly in Desktop and Server CPUs. I want competition all right, but I want it more where it matters.

            • NeelyCam
            • 8 years ago

            No. I want competition everywhere. I’m also predicting that ARM won’t be able to beat Intel in desktop and mobile.

            Those two different things aren’t mutually exclusive.

            • sweatshopking
            • 8 years ago

            I agree 100% with you. I’d like to see x86 working in the mobile space. That being said, i’d prefer windows 8 vs android.

            • srg86
            • 8 years ago

            ARM, x86, MIPS, why not! Even with native code (which I prefer), with good tools you can just recompile and you’re off an away on multiple processors (supplying multiple binaries).

            Must admit I’d rather run a linux distro than windows 8 personally, but ease of running what OS you want, like on a PC would be great.

            • PenGun
            • 8 years ago

            Android’s Linux kernel runs on every damn thing so cooking up a distribution to run on many platforms is quite possible.

            Still the dose runs ARM, or will, so you may get some of your way.

          • ET3D
          • 8 years ago

          I agree that Android doesn’t need x86, but I can definitely see the hacking fun of having an x86 phone. Android is just there to provide a familiar experience, but you can then hack to phone or tablet to run any x86 OS of your choice, which could be very cool.

          • WaltC
          • 8 years ago

          Why people keep on repeating the same old 20-year-old anti-PR for x86 is beyond me. The RISC vs. CISC wars are long over, and the hybrid “R-CISC” won. In short, “x86” today is x86 in name only compared to the state of the instruction set in hardware 20 years ago–x86 today is not “your father’s x86,” etc. Contrary to popular opinion and spin, there is nothing “cleaner” or “purer” or “better” about ARM architectures. BTW, even though some people are only now hearing about ARM, and hence they think it is “new”, the fact is that embedded ARM architectures have been around [url=http://en.wikipedia.org/wiki/ARM_architecture<]since 1987.[/url<] Their claim to fame has always been the embedded, low-power market, with little concern for performance, whereas traditional x86 cpu designs have been primarily concerned with desktop/workstation/server performance with little concern for embedded power profiles. With today's x86 cpu designs you have a billion or two transistors to manufacture at <32nm, and so performance-per-watt has become a key metric in today's x86 cpu design, by necessity. So what do most of the transistors in current x86 cpus do? They're cache, mainly--no "horribly bloated mess" in sight...;) If ARM ever wants to approach x86 levels of cpu performance then ARM, too, will have to add lots and lots of cache. I always enjoy listening to the "horrible bloated mess" crowd ...;) Usually their ideas have been regurgitated from Apple fan magazines of a couple of decades ago, and as such are always entertaining.

            • srg86
            • 8 years ago

            100% Agreed, x86 may have a lot of baggage, but you can hardly call ARM clean, with it’s multitude of instruction sets (ARM, Thumb, Thumb2, VFP, NEON etc), and it has been around for a very long time. It’s had lots of time to evolve.

            R-CISC, or to give it its official name “Post RISC” is the way things are done. x86 processors break instructions down into RISC micro ops and modern RISC instructions sets get larger and larger. The RISC V CISC debate is a relic of 25 years ago.

            As for the amount of bloat that the x86 ISA adds to a processor, these days its less than you think.

            [url<]http://arstechnica.com/old/content/2004/07/pentium-1.ars/2[/url<] In this 2004 article from Ars, The original Pentium needed about 30% of its transistors for x86 backward compatibility, but by the Pentium 4 it was only about 10%. Just think how little it probably is now, even in the Atom.

            • sschaem
            • 8 years ago

            If you take the cache out, its probably ~5% of the transistor count for an ATOM. This would be ~10% of a Arm9.
            But the issue would come from the extra power consumption.

            10% is still bloat. And it gets worse for lower end chips.

            • swaaye
            • 8 years ago

            The ARM ISA has advantages for low power designs. Look at aspects like their Thumb instructions, etc. x86 has a lot of baggage that keeps it from competing with that kind of stuff.

            But with ARM’s CPUs going into the realm of OoOE, 64-bit and such, they are slipping into Intel’s realm more and more. They are going to lose their advantages to x86 simply because they want more speed. And once they start going up against Intel’s world leading manufacturing with the same transistor-eating features, they aren’t going to do very well.

            Intel used to make ARM CPUs less than 10 years ago but sold off the division to Marvell. Still, I wouldn’t be surprised if Intel had experimental R&D in that area just in case.

            (oops srg86 brought some of this up already)

            • Zoomer
            • 8 years ago

            x86 uses CISC instructions on a RISC core. Translating that takes energy. x86 has tons of esoteric instructions from way back that noone really uses regularly anymore, but they still need to be supported, adding hardware complexity and taking up energy. x86 uses variable length opcodes, which is more complex to decode, taking up more energy. All this energy use is unnecessary and can be put to better use elsewhere if the ISA wasn’t x86. Sure, ARM is not perfect, but it doesn’t have baggage dating back to the 8086. Perhaps MIPS?

            Intel might still be competitive, however, because of their manufacturing advantage. Being one process node ahead of everyone else would certainly hide all that nasty power guzzling gremlins really well. But that doesn’t solve the core problem.

            P.S. Apple is a company making stupid overpriced/overprized shit.

            • ET3D
            • 8 years ago

            I don’t think it’s so much an issue of architectures (though it’s of course possible to argue x86 baggage and it’s been done here), but just that it usually makes more sense to stick to a single architecture. Just like Windows for ARM isn’t all that convincing due to software compatibility so is Android for x86. Sure most software will be Java based and run on it fine, but some will be native ARM and will have to be recompiled. That’s not to say it’s a huge hurdle, but it’s still a valid concern for buyers of the “odd” architecture.

            • sschaem
            • 8 years ago

            An ARM9 dual core CPU is ~26million transistor. How much to *ADD* a high performance x86 decoder on top of that?

            The bloat in power and transistor doesn’t matter on 100watt / billion transistor CPU..
            but it still matter for a design that scale all across the mobile spectrum.

            x86 is a bloated mess because its a bloated mess.
            You can see how bad it is if you try to decode IA32/ IA64 + all flavor of SSE/AVX VS ANY risc architecture.

            Claiming x86 is elegant and transistor efficient is pure bull.

            • srg86
            • 8 years ago

            Claiming ARM is also elegant is also pure bull.

      • designerfx
      • 8 years ago

      AMD has always had better graphics performance, so again we’ll have an area that intel will suck.

        • destroy.all.monsters
        • 8 years ago

        On the desktop you’re right. Their mobile chips use Imagination Tech graphics cores just like everyone else (except nVidia – possibly Qualcomm) does.

          • sschaem
          • 8 years ago

          AMD sold off its mobile SoC R&D division to Qualcomm including all IP, designs client contracts (Nokia, Motorola, etc..),
          and thats where Qualcomm Adreno comes from.
          At the time AMD saw no future for itself in the mobile SoC market (even so it was a profitable division of ATI for years with huge contracts) and instead wanted to focused on new designs for netbooks. (Bobcat)

          Recent article that comment a little on this.
          [url<]http://www.tomshardware.com/reviews/touchpad-webos-tablet,2994-12.html[/url<] (That guy was not CEO material.)

Pin It on Pinterest

Share This