45 Comments

Summary:

The new 64-bit A7 chip in the iPhone 5s has been shown to be dual-core, but that didn’t stop it from benchmarking better than the quad-core competition.

A7
photo: Wikipedia

When Apple announced the iPhone 5s last week, it made a lot of noise about the 64-bit architecture of the phone’s new A7 chip. But other than that it remained pretty tight-lipped, which is how Apple tends to handle the specific details of individual components. But we have a better understanding, thanks to the chip review site Anandtech, of how the A7 processor powering the iPhone 5s is getting it done with a 1.3GHz dual-core chip.

You read that right. While many smartphone announcements today can feel like a spec arms race, Apple’s dual-core chip sounds positively quaint. But as we’ve seen demonstrated, more cores do not necessarily equal better performance. This holds especially true for the iPhone 5s, which was shown to handily outperform every quad-core Android phone it went up against.

Anandtech pitted the iPhone 5s against leading Android phones from HTC, LG, Motorola and Samsung. LG currently presents the most competition, as its latest flagship phone, the LG G2, uses Qualcomm’s latest quad-core Snapdragon 800 processor clocked at 2.3GHz per core. While it sounds like that chip should have the upper hand against apple’s 1.3GHz dual-core A7, the benchmarks results told a much different story.

58166

In tests for both CPU and GPU performance, the iPhone 5s beat out every phone it was up against — in most instances by quite a large margin. The testing also included Browsermark 2.0, which I’ve used to benchmark phones in the past. It’s a browser benchmark that gives a good view of overall, real-world phone performance. Again, the iPhone 5s came out on top, if not quite as dramatically as in the other tests.

So what makes Apple’s two cores more effective than so many other phones running four? There’s a fairly simple logic to this. Anandtech explained in its review, “Two cores are still better for most uses than four cores running at lower frequencies. Nvidia forced everyone’s hand in moving to 4 cores earlier than they would’ve liked.”

There’s no doubt that four cores (or more) can make for more effective processing. The trick is in making sure these cores are optimized to do the duty. So while you shouldn’t dismiss specs entirely, it wouldn’t hurt manufacturers to take some time to use the four cores they’ve got to the best of their advantage before jumping the gun and going octa-core.

You’re subscribed! If you like, you can update your settings

  1. Dude, BrowserMark tests your BROWSER, not your HARDWARE. Please get this fact straight before posting misleading BULLSHIT. This is highly unprofessional.
    Higher points in BrowserMark simply means better performance while using the the tested browser app, not the system as a whole.
    Quit your job.

    1. In other words, according to Leo, a great browser can completely elevate crappy hardware. Sweet. We can all forget about updating our devices!

      I hope you’re not in IT.

      Dude.

      1. Scott,

        Leo Li is correct about the browser benchmark. BrowserMark, SunSpider and and the other tests done by Anand (as pointed out below) are all browser benchmarking tools that test the browser, the ratings given do not directly reflect the actual OS abilities. Say you have a crappy browser (say Internet Explorer 8) and you run BrowserMark on it, it’s going to get a very low score, now run the same test in Google Chrome, it will have a very high score. The test runs things like 3D rendering, CSS3 Transitions, HTML5 Canvas, and many other browser specific test. Browsers with Hardware rendering (like Safari for iOS) will do much better than browsers that don’t support it (like Google Chrome for Android). I’ve been researching for a few hours now looking for a decent OS benchmark for the iPhone 5s but I have yet to find one that has reputable sourcing. When the iPhone 5s comes out I would like to see a real thorough set of OS tests done on all the flagship phones before people start making claims about phone performance.

        Have a good day.

        1. Excuse me, “Hardware Acceleration”.

        2. You trying to say anadtech’s test for the 5s is not good enough for you.

          Wow.

    2. You might have read the article too quickly or that the article wasn’t clear about the type of benchmark tested. I am not sure why Browsemark was the only data highlighted, but Anandtech put the the processors though a variety of tests. Here are the link to the data:

      http://www.anandtech.com/show/7335/the-iphone-5s-review/5

      The data shows that the A7 compares well head to head against Intel’s Bay Trail Atom processors and beating it in several categories.

    3. And whats very funny about it is iPhone 5 in youtube videos score around 2300 only !!!
      Check here https://www.youtube.com/watch?v=iATFnXociC4

  2. So, Gigaom talks about how pointless 64-bit processors are for anything but more memory, which I refuted, and now they’re saying how amazing Apple’s processor is. I wasn’t kidding when I said 64-bit makes a big difference, and not just in regards to the amount of memory that can be addressed.

    HOWEVER, all of the benchmarks they ran were essentially single threaded. There were no multithreaded benchmarks, at least, not in the part of the article you seem to be referencing.

    In the multithreaded CPU-testing Physics benchmark, the iPhone 5S was absolutely crushed by the quad core competition: http://images.anandtech.com/graphs/graph7335/58200.png

    A dual core chip can only do so much, but in lightly threaded tasks, a faster dual core chip will always outperform a slower quad core chip. This is obvious.

    1. Ideally, Anandtech would have run several multithreaded tests. Why they did not is inexplicable.

    2. How do you square your statement with the fact that the A7 is clocked at 1.3Ghz while the Snapdragons and Exynos’s are specced at 2Ghz+?

      Apple still has done something pretty impressive neh?

      1. I said it was impressive. The article title is completely, and totally wrong.

    3. technicalconclusions coder543 Friday, September 20, 2013

      What you’ve managed to show is that one small benchmark, clearly a subset of a larger composite benchmark, which appears to contradict the claim in this article. That’s all well and good, but it doesn’t make your claim true. With any difference in architecture, there will always be some small sub-benchmark which deviates for the larger picture.

      You also claimed that no multi-threaded benchmarks were conducted. That’s not true. Geekbench 3 results were published, including the multi-theaded results. Here, The iPhone 5s scored 2564. For reference, the highest scoring Android phone listed on the Geekbench web site is the Samsung Galaxy S4 octo-core variant which scored 2077.

      No matter how you spin this, Apple has clearly retaken the performance lead. At some point within the next year, they’ll likely lose that lead by some margin. Then Apple will answer with the A8, etc. That’s how it goes. The point being, design chips for the mobile market is all about making the right choices and compromises. In practice, betting big on quad core (and higher) chips has not really paid off. Like it or not, there are very few instances where quad core chips are truly leveraged to their maximum potential. Likewise, from a practical perspective, Apple bet big by going with 2 more powerful cores instead of 4 less powerful cores. That decision seems to have paid off for them both in terms of performance and power usage.

      1. According to multiple sources on the internet, the GS4 has a score of over 3,000. I’m still right, the article title is still wrong. Did you not notice how I have repeatedly said the A7 is a great processor? It still doesn’t beat the quad core flagships. In single threaded benchmarks, the number of cores is irrelevant. The title, is wrong.

      2. You stated that very well. It is a see-saw. Psychologically speaking, when people are presented with information that contradicts their belief, they have a tendency to defend the incorrect belief even stronger. It seems like some people are just mad that Apple made a great processor. I’ve got latest versions of Android and a 5S. Both are great pieces of equipment. I look forward to getting the software to use the cores and run 64bit. It’s win win that these Apple/Android compete.

    4. Why is a physics benchmark important? I don’t run folding@home on my phone. When choosing a phone, I look at how fast it runs apps I will likely be running. In this case, the latest Apple and Android phones are both great performers.

  3. Also handily beat the other phones in Geekbench benchmarks.

  4. s3 ranks higher than s4, so lame..enjoy your check from cupertino

  5. Wait for Tegra4 it will take the lead by alot!

    1. technicalconclusions Tim Friday, September 20, 2013

      With an A15 based quad core system, I would expect the Tegra 4 to perform well. It may even beat the A7 at least on the multi-threaded CPU benchmarks. I haven’t seen Geekbench 3 results yet. But, I do know that the GPU that the Tegra 4 is a disappointment. For starters, unlike Apple’s A7, it doesn’t do OpenGL ES 3.0. It’s still a 2.0 chip. Worse, it lags behind the A7 on every GPU test I’ve seen (both on screen and off screen scores). Example… GL Benchmark 2.7 T-Rex: A7 – 37fps (first chip to break 30fps). Tegra 4: 16fps. It lags in 3DMark tests as well. It’s a solid chip overall, but to suggest “it will take the lead by alot” is a bit of a stretch.

  6. Once I get my hands on a iPhone 5S then I will start running BOINC test benches

  7. The Droidboyz will just crank up their quad-core beasts to 2.5 GHz and say Apple is falling behind the hardware curve. There’s no reason Android smartphones should come up short to Apple products. It’s just like Windows PCs vs. Macs. Stick in faster chips than Apple does and the game is quickly over. It should be easy for all the Android smartphone manufacturers to flip a switch to make those quad-cores run faster to outpace the A7. Droid users hate to experience benchmark butt-hurt from iPhones. It’s just a matter of pride. Droid smartphones are meant to rule the world of power users. iPhones are meant for the wuss users.

  8. sure iphone 4s gets a higher score then HTC One.

    *facepalm*

    dumb asses

  9. LOL; Apple could put an abacus in its phones instead of an A7 and the sycophantic press would find a way to say it’s better than the quad cores populating the high-end Android phones. I’m sure a dual-core at a higher frequency will beat a quad-core at a lower frequency on light-threaded tasks. I wonder what happens to that dual-core when the threads get heavy…

    This article reminds me of that old saying about how it’s not the size of this ship but the motion of the ocean. Yeah, sure, keep trying to sell that bull***t. Apple has failed to keep up with the Android world. They won’t put true HD-class screens in their phones. They stay at 4″ screen sizes. They make huge bank because they refuse to put an SD card slot in their phones. Someone from Spain already beat a good deal of their new security features with a simple trick involving the calendar app and an aborted power-off. Finger print sensor? I’d go without a phone before I’d let my fingerprints be digitized and read by who knows what hacker or MIB government agency.

    And yet article after article extols these phones as if they are a gift from the heavens bestowed upon the great unwashed masses. I don’t think I’ve read a single article yet that criticizes anything about these phones. At best, one article has dared question why the 5C has such a high selling price for a supposed entry-level phone.

  10. iPhone5Samesheet Saturday, September 21, 2013

    Galaxy Note 3 is faster and smarter supporting true multi-window multitasking with 3GB DRAM. Search for “galaxy note 3 benchmark gsmarena”.

Comments have been disabled for this post