Skip to main content

Acer Iconia W510 Tablet: A Tale Of Intel Vs. ARM And Acer Vs. Apple

Improving Touch Through Host-Based Acceleration

The next step that enhanced touch reliability and sensitivity involved Intel’s epiphany about gesture detection. Most gestures were picked up using the touch panel's USB controllers with dirt-cheap processor cores. Intel’s performance team developed a secret formula that allowed them to improve the software algorithm for both accuracy and speed by shifting some of the work to the host CPU. Shifting the workload, and then using more sophisticated software, became such a significant innovation that Intel chose not to patent it due to the disclosure requirement. Instead, the company elected to protect this the way Coca Cola's recipe is safeguarded: by keeping it fully secret.

This isn't smoke/mirrors and pure marketing. Under embargo, I spoke with the lead software engineer behind this advancement and discussed the technology behind Intel's optimizations. Nobody from PR was at that meeting to interfere, and I had sufficient information to evaluate and understand the pertinent principles. This one's real. I hate to ask you to "just trust us" on this one, but in this case, you sort of have to. I'm not a full-time programmer, but I still took first place in all of the computer science programming competitions held when I was at Stanford University, both involving fast algorithms and my day-to-day research involving computational biomechanics. So, it's not like Intel is pulling the wool over my eyes. More important, you don't have to take my word or Intel's blindly. You'll know as soon as you try one of these machines at a Microsoft Store.

Currently, this technology is present in the aforementioned systems. Intel is also working with the touchscreen chip manufacturers to ensure more tablets incorporate its advancement. Because the technology requires certain elements of the host processor, it can only be achieved using x86-based devices. Interestingly, even though platforms with ARM processors inside cannot benefit, AMD's x86 cores do gain from Intel's R&D. One of the first touch controller OEMs to license this tech is Elan, and you'll start seeing more x86-based machines with Elan's touch controller sporting Intel's optimizations. There's no easy way for you to identify the devices with performance-tuned touch software without doing some extra research, but we'll keep you updated as we review new tablets and touch-enabled notebooks.

Last, Intel had to optimize its CPU. One of the important factors required for a smooth user experience is a consistent 60 FPS. The company's engineers found that a memory controller optimized for smooth graphics scrolling was different than one optimized for SPEC benchmarks, particularly when a CPU and GPU share the same memory. In fact, it turned out that the best memory controller design for great benchmark numbers was the opposite of what was needed for real-world tasks like swiping across a screen. Do you shoot for a constant 60 FPS and optimal experience, or optimize for industry-standard benchmarks indicative of raw computation in heavy workloads? Intel showed that it was serious about the tablet and phone markets by making what I consider to me the only viable choice (and that's on a site proud of its benchmark-based conclusions). Most of the time, Atom processors go into consumer-oriented devices or micro-servers, rather than environments requiring heavy number crunching. So, the smart move is to optimize for Atom's target markets. Taking a hit on the synthetic benchmark scores, then, Intel went for the gain in usability. It thought like Steve Jobs back in '96 working on the original iPhone: user experience comes first.

Windows 8's software optimizations, Intel's R&D into touch panel optimization, its sharing of knowledge with OEMs like Acer for x86-based devices, and a different memory controller from previous-gen Atom processors mean that today's tablet- and phone-focused Atom SoCs are very different from the ones we poked fun at during the Windows 7 era. The branding is the same, but a system like Acer's W510 with Intel's Atom Z2760 is completely different from the old ExoPC with an Atom N450.The jump in responsiveness that an iPad offers over an Android-powered tablet is the same as what Acer's W510 gives you over an iPad. And Intel gets that performance from a standard PowerVR SGX545.

  • Priox
    "Because shipping was on my tab and only the keyboard was broken, I shipped it on its own."

    You made a massive assumption here to save yourself a few bucks in shipping costs. Your assumption was wrong, and the delay in processing your RMA is all on you.

    Acer manufactures and sells the dock together as a single unit. They separate physically but they are still both part of the same product. It is very reasonable and logical that they would want to examine both together in order to determine the cause of the problem.

    It is not reasonable or logical to compare the Acer W510 dock to a keyboard or mouse for a Mac Pro. Keyboards and mice are not system specific and are highly interchangeable. Your Acer tablet may function without the dock, but the dock does not function without the tablet; it's a system dependent peripheral.

    Next time you make an assumption that turns out to be wrong, I hope you'll accept some responsibility for it.
    Reply
  • ta152h
    What does "Since I never notebooks for repair with their hard drives, I spent time scrubbing my data." mean?

    As if the whining about having to send the full unit in wasn't bad enough (anyone with any technical knowledge would know to send the complete system, instead of being miserly), but then bad English.

    This was a really bad article.

    Reply
  • Bloob
    Not surprising, I have yet to buy anything Acer which works well. The products may just fill the exact function they are advertised to, but not a hair more.
    Reply
  • adgjlsfhk
    so pretty much what I got from this was that
    a. this is a good tablet, much better than an ipad
    b. it crashes if you try to do things that no ipad could ever do (full pc games)
    c. customer service sucks

    don't read the article, this has all the info
    Reply
  • mayankleoboy1
    The Google Octane benchmarks are completely bogus for IE10 VS Chrome on SB-E .

    Chrome currently performs best in Octane. Nearly matched by Firefox. IE10 is nowhere near these two.
    Either you took a busted Canary build, or there is something wrong with the test setup.
    BTW, why are you testing a Canary build here ? Those are very unstable, and perf goes up and down.

    Could you try the latest release Chrome and retest ?
    Reply
  • hons
    so pretty much what I got from this was that
    a. this is a good tablet, much better than an ipad
    b. it crashes if you try to do things that no ipad could ever do (full pc games)
    c. customer service sucks

    don't read the article, this has all the info

    Plus -------- Acer needs him to pay the shipping charge!!!!!!!!
    Reply
  • xtremeways
    I agree, this was a bad article. Acer's customer service is widely known as crap. Nothing new there, but as someone in the IT field you should have already known this. Armed with this knowledge you should have done everything by the book and not assumed you could just send the dock. I rather had read one of those 20 picture articles than this one. Sorry dude.
    Reply
  • duckwithnukes
    Who is this guy? Why the rant? Very poor article.
    Reply
  • Onus
    I just don't get it. So many companies no longer understand, that service is everything. Niche consumers (e.g. do-it-yourself enthusiasts) may not need service, but they're a niche; a tiny percentage of the market. To everyone else, if the product isn't perfect, the service better be, or they'll find something else. Could this explain why so many people are willing to pay the "Apple Tax?"
    Reply
  • ojas
    I have performance numbers for an overclocked six-core PC in there just to remind everyone that the death of the desktop is still a premature proclamation.
    I have a 9 year old 1.7 GHz Single core Pentium M that can prove the same. Sunspider (0.91) score running Chrome (v24) was 544.6 +/- 6%. 1GB DDR RAM, Windows XP, Intel IGP. Don't remember the clocks.

    Sunspider's sensitive to IPC and clock speeds, doesn't seem to care much about core count, as the rest of my little test went like this:

    Core i7-3517U @ 2.4 GHz + Turbo = 208
    Core 2 Quad Q8400 @ 2.77 GHz = 210.3
    Core 2 Duo P8600 @ 2.4 GHz = 262.4

    All within a 2% error margin.
    Reply