ARM Compatibility: Binary Translation

Similar to Apple's move from PowerPC to x86, Intel finds itself in a difficult position with bringing Atom to Android. The OS isn't an issue as it has already been ported to x86 and all further releases will be available in both ARM and x86 flavors. The bigger problem is application compatibility.

There's already support for targeting both ARM and x86 architectures in the Android NDK so anything developed going forward should be ok so long as the developer is aware of x86.

Obviously the first party apps already work on x86, but what about those in the Market?

By default all Android apps run in a VM and are thus processor architecture agnostic. As long as the apps are calling Android libraries that aren't native ARM there, once again, shouldn't be a problem. Where Intel will have a problem is with apps that do call native libraries or apps that are ARM native (e.g. virtually anything CPU intensive like a 3D game).

Intel believes that roughly 75% of all Android apps in the Market don't feature any native ARM code. The remaining 25% are the issue. The presumption is that eventually this will be a non-issue (described above), but what do users of the first x86 Android phones do? Two words: binary translation.

Intel isn't disclosing much about the solution, but by intercepting ARM binaries and translating ARM code to x86 code on the fly during execution Intel is hoping to achieve ~90% app compatibility at launch. Binary translation is typically noticeably slower than running native code, although Intel is unsurprisingly optimistic about the experience on Android. I'm still very skeptical about the overall experience but we'll have to wait and see for ourselves.

 

What's Different This Time Around: Google & A Sweet Reference Platform Final Words
POST A COMMENT

163 Comments

View All Comments

  • french toast - Thursday, January 12, 2012 - link

    No i dont doubt Anands Integrity or his benchmarks, i have just questioned his optimistic assuptions on un proven Intel hardware, especially when Intel produced the scope and the power benchmarks.

    All of which compare last years hardware, all the Intel sorced benches are single threaded and are likely due to Intel software optimisations, and lastly because the atom is higher clocked.
    The benches as exophase has pointed out are on things that the cpu is not loaded/stressed.

    Its great he has provided us with an insight, but this in no way proves that this architecture is the best nor does it prove that it would have 'dominated' android devices last year for a whole number of reasons that i cant be bothered to re-type.

    Anand has jumped the gun on his conclusions thats all im saying
    Reply
  • Dark_Archonis - Tuesday, January 10, 2012 - link

    Time for the haters to eat their words, and the ARM supporters to admit they were wrong.

    On the forums there were enough self-proclaimed "experts" that claimed Intel was doomed and that ARM would take over. Well who's laughing now? Here we see an Atom SoC, that has power characteristics EQUAL or BETTER THAN ARM SoCs, while posting very strong performance as well. This is keeping in mind that it is just the first step for Intel in the smartphone market.

    ARM supporters are already claiming that Medfield only looks good because it's being compared to year-old ARM designs. But guess what, the Atom design is 5 years old! Yes, this is a new SoC for Intel, but the fact is Intel must be given HUGE credit that they were able to make an SoC based on the 5-year old Atom design that is extremely competitive in power and performance to ARM SoCs.

    Next-gen ARM designs will arrive later this year yes, but Intel will not stand still either. Also lets not forget about the next-gen Atom that will come in 2013 on the 22nm process. That will be a HUGE leap forward for Intel and will put them in an extremely competitive position against any quad-core ARM SoCs.
    Reply
  • DigitalFreak - Tuesday, January 10, 2012 - link

    You want a cookie or something? No-one cares about your fanboy rant. Reply
  • Dark_Archonis - Tuesday, January 10, 2012 - link

    Obviously you cared enough to login, and make a useless, childish post devoid of anything meaningful, instead of actually replying to what I had posted.

    Great job at proving yourself wrong.
    Reply
  • name99 - Wednesday, January 11, 2012 - link

    Dude, the reason he is (justifiably) mocking you is that you are comparing ARM silicon that is shipping TODAY with Intel silicon that will be shipping in a YEAR.

    That's how it works among adults --- you compare shipping with shipping, not what I plan to ship (one day) with what you plan to ship (one day).
    Reply
  • guilmon19 - Tuesday, January 10, 2012 - link

    Except that Medfield is built on a 32nm process while ALL available A9s out there are built on the 45nm process. If the A9's were 32nm as well these benchmarks would be very different and these benchmarks came straight from intel, lets way wait till we actually get the phone and do benchmarks from there to determine which one is better. Reply
  • madmilk - Tuesday, January 10, 2012 - link

    Doesn't matter than Medfield is 32nm, by the time anyone else makes a 32nm SoC Intel will be on 22nm.

    It's a massive advantage that can't just be thrown away, especially if trying to make a fair comparison.
    Reply
  • Exophase - Tuesday, January 10, 2012 - link

    Hardly, TSMC 28nm ARM processors are right around the corner while 22nm Atoms won't hit until 2013. Intel's doesn't yet have the same process advantage with their LP SoC nodes. Reply
  • Dark_Archonis - Tuesday, January 10, 2012 - link

    Your point being? Even looking at TSMC's 28nm process, Intel will still have quite an advantage with their 22nm process. ARM will only have an advantage for a few quarters with the 28nm process, then Intel will come out with their 22nm process. You also fail to acknowledge that Atom on 22nm (Silvermont) will a brand new Atom architecture, which you can bet will be a HUGE leap forward in performance and power efficiency.

    TSMC is well known for having lots of trouble moving to new a process. Exophase, you should rephrase "right around the corner" to say in 1-2 quarters at the very least. It will still be a few months before we see any 28nm ARM phones actually on the market.

    Intel has aligned their mobile and desktop process roadmaps, so all future Atoms will move to a new process node as fast as Intel's desktop chips.

    Intel will also have a technological advantage as their 22nm process introduces FinFET, while TSMC won't have FinFET until their 14nm process.

    Furthermore, Intel intends to introduce 14nm in 2013/2014, while TSMC won't introduce 14nm until 2015.

    Intel was really slow getting into the smartphone market, but now that they have their foot in the door, they will not let off the gas. Intel's efforts in the smartphone market will only accelerate from this point. Intel will attack ARM directly at the low-end power scale in this case, and ARM will struggle greatly at the high end as Intel's Haswell will offer revolutionary power/performance ratios. That's not even mentioning what power/performance ratios we will see from the the next-gen Atom design.

    Also let me restate that this is Intel's *first step* into the smartphone market, and they are able to be very competitive with ARM designs.
    Reply
  • Exophase - Tuesday, January 10, 2012 - link

    ... my point being that by the time others have 28nm ARM SoCs - you know, in a couple months - Intel's 22nm Atom will still be a good year off, quite contrary to your original claim.. Did I actually need to type that?

    It's not 1-2 quarters, and we know this because we're seeing Transformer Prime TF202 with Krait, not to mention the release of AMD's Southern Islands which is also on TSMC's 28nm process.

    Of course it'll be a few months before we see Medfield phones on the market too: this article says Q2 for Chinese phones, which means 3+ months, and "by the end of the year" for something released from a decent global player. Not really uplifting schedules.

    Of course Intel certainly may gain a much bigger process lead with Atom on 22nm and beyond, but your original comment was just wrong. That's all I'm saying.
    Reply

Log in

Don't have an account? Sign up now