Comments Locked

185 Comments

Back to Article

  • watzupken - Monday, September 28, 2015 - link

    After going through some of the benchmarks, I think its quite amazing how much the mobile SOCs have improve at this point. It is still quite a bit to catch up with the Intel ULW processors, but they seem to improve by leaps and bounds with each passing generation while Intel is still sitting around idle thinking they are untouchable in the desktop/ laptop arena.
    Anyway, I am impressed with the improvement in the storage and graphics with the new iPhones. Waiting for patiently for a review here. ;)
  • Shadow7037932 - Monday, September 28, 2015 - link

    I'd like to see more competition for Intel in the desktop/laptop space, esp. since AMD isn't doing much these days. With that being said, x86 is deeply engrained in the desktop/laptop space so it's going to take a long long long time for ARM to do any serious impact here.
  • trparky - Monday, September 28, 2015 - link

    I doubt ARM will do anything in the desktop space, a lot of people have legacy software that they must run and ARM basically throws legacy software out the window. There's a reason why Microsoft has decided to make only x86/x64 versions of their Surface tablets now. The ARM versions have no place in the Windows world except for phones and even then that's laughable since Windows Phones are essentially DOA.
  • Pissedoffyouth - Monday, September 28, 2015 - link

    >a lot of people have legacy software that they must run and ARM basically throws legacy software out the window

    Then how come a whole ton of people buy Chromebooks, or use and iPad/Android tablet instead of a PC?

    Business use sure, but average Joe only needs Facebook and Chrome
  • kspirit - Monday, September 28, 2015 - link

    I don't know who this "Average Joe" is who keeps popping up everywhere, but people really do need more than Facebook and Chrome.

    Microsoft Office comes to mind. Mobile version is nowhere near the desktop version, and let's be honest, almost everyone needs Word at one point or another. Google Docs sucks.

    The average student can require anything from Latex to MATLAB to Visual Studio, not to mention the Windows software for students of Economics, etc.. ARM software doesn't cut it. Phones/Tabs and Laptops/Desktops might be getting more stuff in common but one can't replace another. Chromebook isn't even a threat to the Windows laptop market. They're more of a side option than a main PC alternative.
  • uhuznaa - Monday, September 28, 2015 - link

    Well, I think MS Office on a large iPad with a keyboard isn't all that different from MS Office on a laptop or desktop.

    And yes, it's not about a "main PC alternative" -- more and more people really don't need a PC to begin with. I know that this is hard to grasp for techies, but it's still true.
  • Wolfpup - Monday, September 28, 2015 - link

    Nope, it's not. I support a ton of normal users, and not one of them can make do without a real PC. None of them but one use a tablet (and she uses a Surface so that doesn't count) and for all of them a tablet is a luxury while a PC is a necessity. That's no different for me either.
  • aliasfox - Monday, September 28, 2015 - link

    I'm not sure an ambitious high school or college student would be the best gauge of 'average joe-' even an undergraduate business student (like me, so many years ago) required Word, Excel, and Powerpoint for school use.

    However, now, I have Office, SAS, and a variety of database querying tools on my PC at work. And while I have a Mac Pro at home as well as a work issued Dell Latitude, most of my personal activities can be handled by the iPad Air - I'm not creating PowerPoint presentations to show my girlfriend, I'm not pulling thousands of rows of data into Excel to parse, and I'm certainly not writing the Great American Novel (which could certainly be done on an ARM based device, perhaps with a dedicated keyboard).

    Aside from an HTML5 aware browser and access to video streaming services, most people would be happy as clams. The only thing I do on my Mac Pro that I really wouldn't like doing on iPad is photo and video editing, but that's more a function of limited interface options rather than strictly hardware performance and software availability.
  • olde94 - Sunday, November 1, 2015 - link

    Yeah that's right! The show last year of the ipad air 2 handling 8k phots in pixelmator and 4k video, was actually impressive! and thus the interface is the biggest limitation :D!
  • johnnycanadian - Monday, September 28, 2015 - link

    It really depends on the application. We've deployed a custom software solution on iOS to 200+ pipeline inspectors and operators. It's rock-bloody-solid because of the walled garden that is iOS: we don't have to worry about drive-by malware, the staff don't have to update system & antivirus software, and in the cases where they do? Not a single line of code has broken because of a system update.

    iOS & the iPad won't be a panacea for everyone but our lifetime cost-per-node is about 1/5 what it would have been if we'd have deployed standard Windows notebooks (hardware, software & support).
  • blackcrayon - Monday, September 28, 2015 - link

    Thank you. People have no imagination when it comes to the usefulness of a device. Certainly Windows is the swiss army knife of operating systems, but it comes with a ton of baggage in terms of support, maintenance, and malware defense. Sometimes a more appliance-like device is really a better solution. I mean you can easily lock them to a single app over wifi and they'd be virtually unhackable remotely.
  • retrospooty - Tuesday, September 29, 2015 - link

    The point most people make in a scenario like with those 200+ pipeline inspectors and operators with iPads in the field... When people like that get back to the office, they aren't using the iPad, they are using their PC's. It's not that tablets cant be useful, but they certainly aren't enterprise ready and aren't replacing Windows at all... In some cases they are being used in addition to Windows. They are great purpose driven devises though.
  • Wolfpup - Monday, September 28, 2015 - link

    Every normal user I know/support uses WAY more than a Chromebook or Android can provide. And power users, forget it.
  • uhuznaa - Monday, September 28, 2015 - link

    If they need (and get) support, they're not "normal" users.
  • tuxRoller - Monday, September 28, 2015 - link

    Chromebooks are huge in schools.
    Also, floss apps are only a recompile away from arm. You may not use those apps but they are usable to many in Professional Land, let alone being fine for home users.
  • hazydave - Monday, September 28, 2015 - link

    Of course a power user isn't going to claim a tablet or smartphone as their only computer. But a real power user isn't going to be happy on a Surface Pro or pretty much any laptop, either, far as actual PC apps go. Microsoft Office isn't a tool for power users, it's a tool for beancounters and clerks. If you need a powerful wordprocessor or spreadsheet or whatever, there are a dozen to choose from on Android, Office365 included if that's important to you. You can also develop software on an Android tablet... naturally, those things do work better with a physical keyboard.

    Bottom line for me us that anything that's comfortable on my laptop works dandy on a tablet as well... maybe not the exact program, but close enough, and that cones with the things that make tablets good: all-day on a charge, no heat, easy to drag along, paper/magazine density display, etc. I have just as many uses for a tablet that are annoying on a PC as PC things that don't play on a tablet.

    And sure, some of the things that run great on my desktop work uncomfortably on my laptop, but they do work. It can be frustrating on a slow CPU (quad core mobile i7), low ram (16GB), small drive (512GB SSD on PCIe... at least its fast) and just one screen, even at 3840x2160. Some desktop things aren't worth doing or fail completely on the laptop. Natural case of going from any big machine to a smaller one. On a PC tablet, there's no room for that many big applications plus data for them... and few tablet apps designed for mobile.

    More regular users are going to mobile devices as their primary computer every year. I doubt power users ever will... doesn't we we can't find plenty of uses for what tablets actually do well.
  • vampyren - Friday, October 2, 2015 - link

    Totally agree, i would never be able to replace my laptop with a tablet. Also i hate the flanky tablet keyboard.
  • olde94 - Sunday, November 1, 2015 - link

    I have a latex app on my ipad, and i use sharelatex on my ipad if i don't have anything else handy ;).
    I've tried code java and python on my android phone, and as a mechanical student, most of those math script run just as fine as on matlab. (it was test's for fun)
    I know math software like matlab and maple ar working on arm versions. i think matlab said something about 2018 was a esimat for offline matlab on IOS. (not sure though)
    If we count FLOPS performance the new z5 compact delivers more flops than a asus ux305, which is "good enough" for most. (i know one with one on which i have tested stuff on ;) )
    I don't know the performance on a ipad pro, but we are SURELY gaining on intel! at FAST pace.
    I had a surface 1 tegra 3 processor, and (aside from the SoC being too slow) i saw only one flaw. App's had to go through the store so java was never available = no java software = a lot of software lost.
    i give my setup 10 years. in 10 years i think i will run my pc off an ARM SoC. And i would consider myself a heavy user with CAD (Through a VM), Scripting, VM's, 3 screen setup on my macbook pro (13" (at home)) photo edit, and general consumer ;)

    ARM is capable of more than you think ;)
  • nikon133 - Tuesday, September 29, 2015 - link

    More, like, Average Joe's grandparents... If Joe needs no more than browser, he will find himself quite a bit below average today, imho.
  • Michael Bay - Monday, September 28, 2015 - link

    I quite like my "DOA" WP and see a lot around.
    Not to mention ARM was always a very viable platform for Windows PoS implementations.
  • nikon133 - Tuesday, September 29, 2015 - link

    Windows phone is definitely niche, but I wouldn't call them DOA. They were/are reasonably popular in Europe - in some countries not trailing much behind iPhone (Italy comes to mind first). Overall, they hold as much as 50% of iPhone's market share in EU5 (GB, Germany, Italy, France and... Spain, I think?). Of course android is dwarfing both of them, but still.

    What is interesting is that they do much worst in US, which should be home turf for them. Might be down to Nokia not being too popular in US, while still being considered Euro brand in Europe (even if it is really Microsoft behind them) and commanding some loyalty based on that... speaking of which, it will be interesting to see if brand transition from Nokia to Microsoft will introduce any significant changes in WP market-share, both US and EU...
  • Azurael - Wednesday, September 30, 2015 - link

    I don't really understand how WP has done quite well in Europe - I wish my Lumia 930 ran Android! I prefer it to my Nexus 5 or G2 from a hardware point of view. But WP is a mess. Despite being weeks from release, WP10 still feels like an early alpha with its kludgy animations, crashing built-in apps and UI concepts like hamburger menus so badly copied from Android. I actually prefer the much cleaner UI in WP8.1 but it's missing half the functionality you expect from a modern smartphone and just the settings menu would give people with OCD recurring nightmares.
  • name99 - Monday, September 28, 2015 - link

    A9X has caught up with Intel ULV for most practical purposes.
    Look at my comparison of A9 with Broadwell-Y
    https://browser.primatelabs.com/geekbench3/compare...
    Broadwell-Y is ahead, but barely.
    My guess is that A9X will zip past Broadwell-Y and put up a good show against Broadwell-U when I run the comparison of iPad Pro with a MacBook AIr.
  • Morawka - Monday, September 28, 2015 - link

    broadwell is old, try comparing to skylake please.
  • name99 - Monday, September 28, 2015 - link

    (a) You think there's a serious difference between Broadwell and Skylake? Good for you.

    (b) Apple doesn't OFFER a Skylake Macbook.
    The damn Skylake Core-M was only released a few weeks ago. I can't find a single review with numbers on the web --- which suggests that manufacturers don't exactly think it's enough of an improvement to immediately swap into their current Broadwell designs...
    The ONLY numbers I can find are :
    Core M-5Y71 2x 1.20 GHz (2.90 GHz) HT: single 2999, multi 4833
    That's basically just the same as the Broadwell numbers.
    The Apple i7 model will run at 1.3GHz, turbo to 3.1, and those enumbers may be a few percent higher --- doesn't change the big picture.

    But sure, sure, the world will be turned upside by Koby Lake. That's the one that's REALLY going to
    show Intel's awesome skills, right? We've been hearing that since Sandy Bridge --- who knows, maybe one day it'll even be true...
  • tmr3 - Tuesday, September 29, 2015 - link

    Those numbers for that Core M processor seem basically the same as the Broadwell numbers because the Core M-5Y71 is a Broadwell chip, not a Skylake chip. Easiest way to tell the difference is that Broadwell-based Core M designs are designated Core M-5Yxx, while newer Skylake-based Core M chips are branded Core m3, Core m5 and Core m7.

    Additionally Geekbench isn't the best indicator of relative performance across different architectures.
  • id4andrei - Monday, September 28, 2015 - link

    GB is irrelevant across ISAs. It's inconsistent even between ARM platforms. Anandtech does not even reference it. They use if for ios to ios comparisons. Try again next time.
  • clemsyn - Monday, September 28, 2015 - link

    I definitely agree with this. Geekbench alone is not a legit comparison for x86. Just look at the source of Geekbench and you will see why.
  • DERSS - Monday, September 28, 2015 - link

    This is why this article provides a lot of other tests where A9 is the first ARM ISA SoC to ever come close to Intel.

    A9X, no doubt, will catch up to Intel on all major tests, whichever you would throw at it.

    But, as others said, this has nothing to do with laptops since Apple does not really plan to replace Intel with their ARM SoCs in the future -- makes no sense since Intel is efficient enough and does not cost that much for Apple (Intel has unique, special prices for Apple).
  • NEDM64 - Monday, September 28, 2015 - link

    because.......

    The tests are the same.
  • larry6hi5 - Monday, September 28, 2015 - link

    I'm super excited to get a 6s. Thanks for the quick preliminary review!
  • Tech_guy - Monday, September 28, 2015 - link

    me too mine's on the way!! I can't believe the performance increase, it's INSANE considering iPhone 6 was already the smoothest gaming device.
  • Gondalf - Monday, September 28, 2015 - link

    After all these are only benches dear Tech :), your nick tells you that a massive data/code caching could do miracles :). It is not all gold but there is even a little of silver here
  • Tech_guy - Monday, September 28, 2015 - link

    the benchmarks can't all be faked. Especially the GFXbench, that's what I care about, 3D performance eats the S6 Edge alive!! Especially since it uses an overkill Quad HD display, it hurts frames per seconds in games. Apple is the winner for me.
  • Morawka - Monday, September 28, 2015 - link

    and that would only matter if iOS games had customizable settings and scablable frame-rates.. All iOS games are frame-rate capped, so new GPU's have no effect on existing games.

    Meanwhile on a PC, if you upgraded the GPU, it would give you better settings, and better FPS across all games no matter how old.
  • blackcrayon - Monday, September 28, 2015 - link

    iOS games are often updated to take advantage of new SoCs with higher quality visuals. Sure you have to wait for the developer to do it, but it happens.
  • DERSS - Monday, September 28, 2015 - link

    FPS on existing heavy CG iOS games is increased automatically; no updates even needed (though they are welcome if developers want to add additional effects).
  • Tech_guy - Friday, October 2, 2015 - link

    iPhone 6s is the best gaming device by far. It's in another league from everything, PS Vita and 3DS watch out this thing is monstrously beastly.
  • darkchazz - Monday, September 28, 2015 - link

    These are some impressive results...

    I hope Qualcomm can deliver comparable performance with their new Hydra CPU.
  • Tech_guy - Monday, September 28, 2015 - link

    seriously these benchmarks probably keep Samsung and Qualcomm engineers up at night lol. But yes it'll be interesting to see.
  • Gondalf - Monday, September 28, 2015 - link

    But no :). This SOC has 8MB of L3 and 3MB of L2, so it is all data and code caching with these short benchmarks suited for SOCs with relatively little caches.
    The real world usage will be a lot slower, moreover likely all this is not sustained like is happened with A7.
    After all it is only a 3W SOC there isn't a single chance it is two or three times faster of 7420 3W too and done on the same process.
    Samsung and Qualcomm are pretty quiet trust me, there is a lot of benchmark optimization in this device. The gigantic caches are there to fake all the phone cpu benches.
  • Tech_guy - Monday, September 28, 2015 - link

    The performance and smoothness in games I get on my iPhone 6(yes the 6) vs my girl friend's S6 makes a strong case that these benchmarks are NOT inflated or faked. I totally believe these as truth, the 6s is the faster device. I mean how can you even argue 2500 single core geekbench vs 1450 single core on the 7420. That can't be "faked" Geekbench is a neutral test they don't care, it's totally based on HOW FAST the cpu's actually can complete the tests.
  • Gondalf - Monday, September 28, 2015 - link

    You ""believe"" firmly, still you don't read the posts of other writers.
    My point is about the massive caching (9MB of cache) that gives massive advantages on very small phone benches. Do you have undestood??

    And no Geek is not neutral, it has massive issues as the cache grow up.
  • name99 - Monday, September 28, 2015 - link

    IF your claims were true --- that all it takes to make a fast low-power CPU is to dump 9MB (don't know where you get that number from -- it's 8MiB L3 and 3MiB L2) of cache on the die, then why didn't Apple do that last year, or the year before? Hell what's to stop Samsung doing it now?
  • Morawka - Monday, September 28, 2015 - link

    That L3 8MB is actually the SRAM which is used by fingerprint device and other devices.. The L2 is what the cpu and gpu have dedicated access to.
  • name99 - Monday, September 28, 2015 - link

    Some day, dude, you're going to look at this stage of your commenting on AnandTech and cringe...
    To give just one example:

    Look at the graph at the bottom of this page:
    http://www.anandtech.com/show/8554/the-iphone-6-re...
    If the CPU has no access to the L3, what do you think that plateau between 1024 and 4096 refers to?
  • tuxRoller - Monday, September 28, 2015 - link

    Why? Why would it need that?
    SRAM for that? Why? Trustzone (it had its own memory system) is used for this sort of thing and it's perfect for it.
    An interrupt causes trustzone to take over, it handles the data and responds, one way or the other, with ios kernel never touching the data.
  • tipoo - Monday, September 28, 2015 - link

    Not saying it's faked, I completely believe Apples single core performance lead, but since you mentioned GB specifically I think Anandtech had some qualms a while ago about discrepancies between the iOS and Android builds of it. Hardware decryption also throws a very large number into the averaging. That's why you may see ARM chips matching Core M, but then in something like Linpack the Core M is 8x faster.
  • name99 - Monday, September 28, 2015 - link

    The Geekbench3 workloads ARE smaller than their desktop workloads. This is true; the sizes of the data sets for each workload are available at www.primatelabs.com/geekbench3/workloads.pdf.
    This is because (in the past) the cores were slow enough that using full workloads would have taken forever. BUT

    (a) John Poole says he has run the full workloads on various mobile cores and in all it makes less than 1% difference.
    (b) Geekbench 4 will come out soonish, and will use identical workloads for mobile and desktop. John has said nothing about this, but I'm guessing it doesn't change the ratings much between x86 and various ARMs --- if it did he'd be preparing the way, I suspect, warning people not to take Geekbench 3 too serious anymore.
    (c) These devices ARE running real world code --- what do you think browsers are --- and the performance numbers behave in pretty much the same way for all the benchmarks.
  • Morawka - Monday, September 28, 2015 - link

    because on the geekbench benchmark, it is never writing to main memory on the 6S, it's using fast cache all the time.. I agree, the cache sizes are making this soc look waaay better than it actually is. in real world usage, your app is gonna use way more than 8MB of that cache, therefore requiring access to main memory.
  • tuxRoller - Monday, September 28, 2015 - link

    Can you cite an example of proof that Apple has ever gamed benchmarks?
    There nothing wrong with use a large amount of cache. Intel's cache implementation is a responsible for a decent amount of their performance.
  • DERSS - Monday, September 28, 2015 - link

    Not really; tests do everything in their power to go above caches, including 3 MB and 8 MB (Sun Spider does not count).
  • lilmoe - Monday, September 28, 2015 - link

    Geekbench is a really bad cross-platform benchmark. Just like browser benchmarks, it's only good for test for generational upgrades of the same platform, and even that has its caveats.

    Heck, forget caching, even the workload isn't equal among different ARM based platforms.
  • lilmoe - Monday, September 28, 2015 - link

    edit: **It's only good for testing generational upgrades...
  • NetMage - Monday, September 28, 2015 - link

    Actually Ars says the A9 is much better than previous Apple chips at sustained (and of course better than all the competition).
  • xype - Tuesday, September 29, 2015 - link

    Gondalf, the fact that you write "The gigantic caches are there to fake all the phone cpu benches." just shows that you have absolutely no fucking idea what you’re talking about.

    If you honestly believe Apple does _anything_, _ever_ to "fake all the benches" you simply have no idea how Apple works, and neither do you have the faintest idea of how Apple’s customers decide on which hardware to buy.

    Just because tech sites like to use benchmarks to geek off about hardware that doesn’t mean than anyone in the real world gives a shit about them. For 99% of the people, the phone has to run smooth and play games with cool graphics. As long as Apple can deliver that (and deliver they do, no matter much deep into the sand you try to bury your head), noone gives a rat’s ass about benchmarks. And I’m pretty sure the same is true for most Android/Windows Phone users.
  • Wolfpup - Monday, September 28, 2015 - link

    Qualcomm's CPUs have yet to impress, so I wouldn't hold my breath. They talk big, and always release things that are worse than whatever ARM's doing, making the whole thing seem pointless. Happened with Snapdragon "it's as good as A9!", happened again with Krait.

    They get design wins because they're "decent enough" and have a modem integrated, not because they're cutting edge.
  • Pissedoffyouth - Tuesday, September 29, 2015 - link

    uh what? The SD800 was a very impressive chip when it was released...
  • retrospooty - Tuesday, September 29, 2015 - link

    Yeah, I dont know what wofpup is talking about. Krait's were awesome in thier day and always did better than standard ARM desings out at the time... The issue with Qualcomm is after the amazing SD800, they didn't have a follow up ready fast enough. The 801 and 805 are just iterations of the 800. Of course they didnt have their 64 bit ready so they went with a stock ARM design on the SD810 - burn. The SD820 looks like another winner though.
  • Drumsticks - Monday, September 28, 2015 - link

    I love how Anandtech's pipeline articles are basically as good as the main articles for other websites. Thanks for the info!

    While I'm not really an iPhone guy - little things about iOS bug me, which is purely IMO - Apple's hardware team has continued to impress since the A6. I do admit loving what they turn out in terms of performance and ingenuity. I'll look forward to the final review!
  • maecenas - Monday, September 28, 2015 - link

    Agreed, even if you don't purchase these devices, Apple's advances push everyone else in the industry.
  • ddriver - Monday, September 28, 2015 - link

    LOL, yeah right. Because apple is not a desperate copier the last few years with barely anything new, and still lagging behind in many areas. I really mean it ;)
  • III-V - Monday, September 28, 2015 - link

    They're ahead in some areas, and behind in others. So is everybody else.
  • melgross - Monday, September 28, 2015 - link

    Oh, jeez! There's always someone to mouth this crap.
  • tipoo - Monday, September 28, 2015 - link

    Just look at the article, man. Even fanboys can't deny their SoC chops.
  • ddriver - Monday, September 28, 2015 - link

    That would barely materialize in any new benefit. Putting more and more powerful chips into useless toys is the lamest possible thing to do. Today people are the product, monetized through the devices they purchase, ultimately resulting in consumers being used through those devices more than they use the devices.
  • blackcrayon - Monday, September 28, 2015 - link

    The only thing missing from your post is a link to how your wife makes $500/day without leaving home.
  • ddriver - Tuesday, September 29, 2015 - link

    Sorry, I doubt you will make 500$ anytime soon, unless you sell your organs or something... but that will be a one time deal ;)
  • cpugod - Tuesday, October 13, 2015 - link

    @ddriver - By pulling your thing out of pants and peeing all over you're just showing us how small it is (and being a douche).
  • KPOM - Tuesday, September 29, 2015 - link

    Only $500/day? She needs to start getting into timeshares.
  • Hemlocke - Tuesday, September 29, 2015 - link

    "Today, people are the product, monetized through the Android and Windows devices they purchase..."
    FTFY
  • watzupken - Monday, September 28, 2015 - link

    Yeah, I have to agree that Apple seems to be one of the main driving force here. Being the first to release a 64bit SOC, it caught many by surprise and now was followed by a mad rush to 64bit SOCs in the Android world. The industry seems to peg Apple as the one to beat with their yearly product update cycle.
  • KPOM - Tuesday, September 29, 2015 - link

    Exactly. Remember how 64-bit was just a gimmick and useless without 4+GB of RAM? Now everyone has 64-bit chips. It will be the same with NVMe.
  • Laxaa - Monday, September 28, 2015 - link

    Exciting. I'm in the market for a new phone this year, and I'm eagerly awaiting reviews of the fall flagships before I make a purchase.
  • sprockkets - Monday, September 28, 2015 - link

    "However, it’s often difficult to understand exactly how much has changed with an S model as Apple tends to focus on high level features, this despite the fact that so many of the changes in an S model are at a low level."

    Oh that's ok, because "The only thing that's changed is everything." /s
  • Infy2 - Monday, September 28, 2015 - link

    Seems to me that this Twister SoC matches or even exceeds perf/W anything Intel has to offer.
  • kspirit - Monday, September 28, 2015 - link

    Both processors are running different software. Not a fair comparison. Windows may well be less power efficient than iOS. Doesn't mean Intel processors getting beat by a phone SoC in raw compute power.
  • melgross - Monday, September 28, 2015 - link

    Well, here's some testing from today's iphone E 6s review. It seems that these SoCs are better than what one might think. It will be really interesting to see how the iPad Pro coming out in November will do. The iPads usually have three cores rather than two, and are clocked just a bit higher. We do t know what the difference between the A9 and the A9+ will definately be, but as a minimum, that will be the difference. Sometimes, there are more GPU cores as well.

    But this is what we see today for the iPhone 6s A9:

    http://cdn.arstechnica.net/wp-content/uploads/2015...
  • Gondalf - Monday, September 28, 2015 - link

    He he, you know that Geekbench dataset for phones is 8MB......so these are fake results :), all the bench runs on the SOC, the code in the L2, the dataset in the L3.
    Apple has stolen the bench to be better than competitors :))).......Samsung included ummmm
  • melgross - Monday, September 28, 2015 - link

    Oh please. That's total bs. I hope you're just joking.
  • tipoo - Monday, September 28, 2015 - link

    Difference in dataset size affect run times, not results. Primate said in their tests there was less than 1% variance.
  • blackcrayon - Monday, September 28, 2015 - link

    You have a problem with Geekbench, which this article doesn't even mention, but somehow can't explain the great results the A9 gets in the several tests that the article *does* mention... Nice.
  • Gondalf - Monday, September 28, 2015 - link

    I have forgot to tell you that GeekBench is able to recognize a Pc core (Haswell, Piledriver etc etc).
    When a Pc core is up Geek shift the dataset to 32MB, giving an HUGE disasvantage to the x86 core under test :).
    So these are FAKE results and you can't do any comparison :).
  • melgross - Monday, September 28, 2015 - link

    More bs.
  • tipoo - Monday, September 28, 2015 - link

    If what you are saying was true, Crystalwell would post godly Geekbench results with it's 128MB L4 cache.
  • name99 - Monday, September 28, 2015 - link

    Oh for fsck's sake. At some point grow up and accept reality.
    https://browser.primatelabs.com/geekbench3/compare...
  • id4andrei - Tuesday, September 29, 2015 - link

    Don't you understand that Anandtech ignores GB for pretty much anything? Why are you clinging so hard to GB? It's irrelevant.
  • Shiitaki - Thursday, October 1, 2015 - link

    Not fair? You are right! The Apple SoC is in a phone! The Intel chip is in a much larger tablet, with a much larger TDP! It is like when the burned out Chevy pulls up next to the Ferrari, and looses by a car length. The fact that they are close is the big deal. The fact that Apple did not spend as much for their processor as Microsoft did in the surface is a bigger deal for Intel.
  • eason000 - Monday, September 28, 2015 - link

    Many thanks to this fresh report! Have be eager to know this for a long time!
    Although I have just signed in for submitting this comment, I have browsing frequently at the techweb for over a year. This place is quite knowledgable, If I may say that, every article seems rigorous and precise. Keep moving on, AnandTech!
  • Tech_guy - Monday, September 28, 2015 - link

    Wow. I knew some of these benchmarks already, but seeing them in the charts like these are just incredible. Apple seems to use the most quality high performance components as seen in the Nand peformance. I mean that's just incredible, beats the crap out of Android. I have an Android because I have to tether it with Sprint (rooted custom tether software etc). But I would drop Android in a second if I could and use 6s as my daily driver.
  • Tech_guy - Monday, September 28, 2015 - link

    The interesting thing to me is iPhone 6s is considerably more graphically capable vs it's larger brother 6s Plus. They're using the same exact GPU apparently (no overclock on the 6s Plus this time). 6s Plus is burdened with 1080p of course, 6s ~750p

    GFXbench Manhattan onscreen. open gl
    6s Plus 38.4
    6s 55.9

    6s around 17 fps better score. 38 is still excellent, but dang that's quite the gap. I'm a gamer and iPhone 6 was always smoother than 6 Plus in games. This is why I purchased 6s, and I still think 5.5" is too large for my average sized hands haha.

    Anyone else surprised Apple didn't bump up the 6s Plus' GPU at all?
  • protomech - Monday, September 28, 2015 - link

    Onscreen video tests run faster on the 6s because the screen resolution is lower. Look at the offscreen tests for a better device-to-device comparison.
  • Tech_guy - Monday, September 28, 2015 - link

    yes this is what I'm saying. They use the same GPU but iPhone 6s Plus has to run the larger resolution, so therefore iPhone 6s is faster.
  • lilmoe - Monday, September 28, 2015 - link

    Not necessarily. Most games don't render at native screen resolutions.

    Offscreen scores is what you should be looking at.
  • Morawka - Monday, September 28, 2015 - link

    offscreen is good for showing how much it can stretch it's legs but at the end of the day, the device is still burdened by the resolution of the screen it uses, so that point is moot. if it runs like shit onscreen, it will run like shit in real world usage.
  • blackcrayon - Monday, September 28, 2015 - link

    Run like shit or look like shit- some games on Android actually render at lower resolutions and scale to the display- 1440p is still often a bit much for a complex 3d game to handle on a phone.
  • melgross - Monday, September 28, 2015 - link

    The reason is that the 6s uses two times assets, as Apple calls it, and displays them directly. The 6+ and 6s+ use three times assets, but then has to Rez them down for the screens higher Rez, which isn't enough for the full three times asset. That takes an extra step, and results in slower on screen graphics performance. I'd bet that if Apple has a three time Rez screen, rather than one thata about 1.5 times the Rez of th )s, graphs would be about as fast as that of the 6s.

    Next year will be the year that Apple does something with the screen. It will be interesting to see if they bump the 7 to 1920x1080, and the 7+ to 2208 X 1242, which would exactly match the three times graphics for the larger phone.
  • MikhailT - Monday, September 28, 2015 - link

    Doesn't Apple have a hardware scaler built into the chip to handle the scaling up and down pretty efficiently? In other words, scaling down from 3x assets to the screen panel might not be a big hit, maybe less than 5%.
  • NetMage - Monday, September 28, 2015 - link

    But it is still drawing to a 2208x1242 screen before scaling down to 1080p.
  • Morawka - Monday, September 28, 2015 - link

    the 6S Plus UI runs at 30 FPS almost always,, meanwhile the regular 6S has a smooth 60FPS UI
  • msae - Wednesday, September 30, 2015 - link

    Do you have a source on that, or have both phones to compare? I have the 6s Plus, and I'm pretty sure there is some stutter in the UI, notably when scrolling in image-heavy apps like Pinterest. Also when launching 3D touch menus, the animation doesn't seem as smooth. Can anyone with a 6s see if this is common? I'd like to know if it's an iOS 9 issue, or if the 6s Plus is as gimped as the 6 Plus.
  • NetMage - Monday, September 28, 2015 - link

    I believe that is because the Plus actually downscales to display 1080p so onscreen it is generating 1242x2208.
  • vagabondssigh - Monday, September 28, 2015 - link

    I wonder why the s6 internal NAND performance grade is so low, when I use androbench on my s6, I got sequential read/write 320MB/S,144MB/S, random 4K read/write 79MB/s,20MB/s, I wonder how you guys got the grades in the diagram.And S6 gets much higher grades using sumsang's web browser in the the krane, google octane and webXPRT tests.
  • lilmoe - Monday, September 28, 2015 - link

    I've pointed that out before. I'm glad someone else finds that strange, because it is.
  • Andrei Frumusanu - Monday, September 28, 2015 - link

    Remember to set buffer to 256K and threads to 1.
  • Kuzi - Tuesday, September 29, 2015 - link

    What is the reason to do that? To make Apples numbers look better?
  • kogaharukka - Thursday, October 1, 2015 - link

    I wonder why Anandtech reviewer used androbench 3.6 instead of androbench 4.0 for android phones, since androbench 3.6 does not show real NAND performance and gives very low numbers.
    The latest updated androbench version is 4.0. Androbench 4.0 is much better for real NAND performance check. NAND performance numbers in the review needs to be corrected with up-to-dated benchmark.
  • UltraWide - Monday, September 28, 2015 - link

    Can you run the benchmarks with full disk encryption turned on? Thanks!
  • tipoo - Monday, September 28, 2015 - link

    They're on by default on iOS, and I don't think you can even turn it off.
  • UltraWide - Monday, September 28, 2015 - link

    Didn't know that, thanks!
    It's even more impressive if encryption is indeed turned on!
  • tipoo - Monday, September 28, 2015 - link

    Since it's hardware accelerated you wouldn't see much of a dip anyways
  • uhuznaa - Monday, September 28, 2015 - link

    Exactly, and it even wouldn't matter since the encryption is done in hardware. But I guess UltraWide meant the Android benchmarks. Would be in fact interesting, I think the Nexus 6 had encryption enabled (for some reason it doesn't use the hardware encryption available on the SoC) which makes it so damn slow in I/O.
  • Pissedoffyouth - Monday, September 28, 2015 - link

    >for some reason it doesn't use the hardware encryption available on the SoC

    It's slower to use Qualcoms HW solution than software solution
  • uhuznaa - Monday, September 28, 2015 - link

    Wow. Even slower than THAT slow (look at the benchmarks)? Or was this without encryption? I'm a bit confused here, granted.
  • tipoo - Monday, September 28, 2015 - link

    Am I nuts or does it look suspiciously like the "driver overhead test" is just measuring per-core performance?
  • jameskatt - Monday, September 28, 2015 - link

    Despite the improvement in CPU and GPU speed and power, they still are a long way away from doing real desktop tasks.

    For example, iPhone and iPads are optimized for single-small screen use only.
    But on my MacBook Pro 15, I use 4 screens simultaneously.

    The iPhone and iPad are optimized for single-app use only - even when multitasking, it is limited by the single screen model.
    On my Mac, I use 15 apps simultaneously - even using 3 browsers simultaneously. And even then, it is still too slow for my taste.

    Memory and storage also on mobile devices is tiny tiny tiny.
    On the Mac, 16 GB is tiny.
    On the Mac, I'm connected to 7 1-5TB hard drives, yet that isn't enough storage - not even considering the backups involved.

    As Steve Jobs said, some people need a truck. That is what a desktop/laptop is for. And mobile devices are simply nowhere near their power nor storage capabilities. Sure, if single tasks are all that is done, one can get by. But when you use more then they fail.

    Try to do Photoshop on a 100 GB file on an iPad. Hmmm, it doesn't work.
  • iwod - Monday, September 28, 2015 - link

    I think you first learn to learn the different between Hardware and software limitation.
  • MikhailT - Monday, September 28, 2015 - link

    What is your point? Nobody said to get an iPhone or any other smartphone to replace your desktop, WTF?

    These aren't meant to replace your desktop tasks, that's why it is called desktop tasks and not smartphone/tablet tasks. I rather have a laptop or desktop that does best with certain tasks than one device that does shitty tasks at everything, which is why I have an iPad in addition to my rMBP. At work, I do everything on my laptop but at home, I do everything on the iPad with no problems. The iPhone and/or iPad does certain things best (content consumption) and for many people, that's all they want. I have no needs for my iPad to do everything the laptop can do.

    On iPad Air 2 with iOS 9, I'm able to do most of my stuff in the split view with two full apps that works out pretty damn well. I'm actually looking forward to getting iPad Pro, so I can more content in two full iPad apps.
  • Morawka - Monday, September 28, 2015 - link

    Please add regular 6S battery comparison
  • NetMage - Monday, September 28, 2015 - link

    I'm sure it will be in the full review. This is just preliminary results.
  • iwod - Monday, September 28, 2015 - link

    Lots of questions in my head.

    Random Read Write is slow, since we can't get enough NAND doing parallel work, we are not likely to get any more improvement. ( But the number still seems suspiciously low for me )

    How likely are we going to see Xpoint instead of NAND. Solving the need for SLC cache and Random Read write speed problems. ( BTW is there 64GB TLC + xGB SLC within? if so how many GB for SLC )

    Or will we see iOS gets the introduction of Page files, with Xpoint used as pages? So 2GB RAM and 2GB of Xpoint. Page file or not, Xpoint would help as we don't need sophisticated controller for NAND speed, saving energy. And getting 2GB extra means updating iOS no longer requires cleaning your iPhone every now and then.

    8MB Cache. I still remember some years ago Anand pointed out SRAM is energy expensive, have we solved that? It seems with every node apple will double the SRAM as well. Most computer on earth don't even have 8MB Cache. So could we expect 16MB when we reach 10nm ( That will be iPhone 7s BTW ). And if power requirement isn't stopping us, will we likely see 2.5D SRAM layered on with the whole layer dedicated to SRAM, getting 32 or even 64MB SRAM isn't far fetched. ( Cost issues aside, but cost isn't as much of a concern for Apple, since they sell whole devices, not Chips )

    Ars suggested the GPU is PowerVR 7 ( Whoa, Apple is moving fast ), are we bandwidth limited yet? If so, when will we see Wide I/O 2 being deployed? iPhone 7s ? Will we see 4GB then?
  • tipoo - Monday, September 28, 2015 - link

    And if the random read/write is due to the testing limitation, what's with the other phones doing better regardless?
  • Hemlocke - Thursday, October 1, 2015 - link

    They do better because the limitation of the test polling (1) is the same limitation on their storage clusters for random R/W. If Apple is using a laptop/desktop class NAND controller, they have no such limitation, and multiple clusters could perform R/W, but the test won't measure anything beyond the standard.
  • tipoo - Monday, September 28, 2015 - link

    I went over the article again, but I can't seem to see where they say it has 8MB SRAM? I thought Chipworks said 4.
  • tipoo - Monday, September 28, 2015 - link

    Oh, nevermind, found it, Chipworks meant each of two SRAM blocks was 4MB.
  • milli - Monday, September 28, 2015 - link

    'In 3DMark, we see the continuation of a long-running trend in the physics test in which the primary determinant of performance is clock speed and memory performance as data dependencies mean that much of the CPU’s out of order execution assets go unused.'

    I don't agree. I would say, it's much more dependant on the total cpu performance which is:
    core-performance x core-count
    That's why Apple scores bad with their dual core cpu's. Not because the out of order execution assets go unused, which is not true.
  • defferoo - Monday, September 28, 2015 - link

    take a look at FutureMark's analysis of iPhone physics tests and you'll understand

    http://www.futuremark.com/pressreleases/understand...
  • milli - Monday, September 28, 2015 - link

    That article is just a case study with a narrow scope. It just shows that the A7 chip hits other bottlenecks which don't allow it to perform faster than the A6 core. Doesn't change the fact that the iPhone has a bad physics score because it's only dual-core. 3DMark's physics test is multithreaded and it can take advantage of even 8 cores.
  • defferoo - Monday, September 28, 2015 - link

    the fact that the ipad air 2 performs about the same as the iphone 6 with an additional core should already put to rest the "core count" fixes everything argument. also, that article is specifically analyzing the test that you're talking about, I don't know what you mean by "just a case study with narrow scope".

    for reference, ipad air 2 review - http://images.anandtech.com/graphs/graph8666/69027...
  • blackcrayon - Monday, September 28, 2015 - link

    That doesn't appear to totally be the case. If it were you would expect the iPad Air 2 to be further ahead of the iPhone 6.
  • tipoo - Monday, September 28, 2015 - link

    I have a question, if the low random read/write scores are due to a testing limitation, why does Samsung and even lesser Android phones do so much better in the same test?
  • mabellon - Monday, September 28, 2015 - link

    Awesome preliminary. The comparison with the Surface Pro 3 is nice context. However that is the Core i5 model with 15W TDP, and Turbos even higher. Do you think you could include the new Macbook 12" and the Surface Pro 3 (Core i3) as reference? Their Y-series parts are perhaps a better comparison. Can't wait to see the inevitable Surface Pro 4 vs iPad Pro performance comparison. Competition is great! Thanks!
  • name99 - Monday, September 28, 2015 - link

    https://browser.primatelabs.com/geekbench3/compare...
  • Chaser - Monday, September 28, 2015 - link

    Apple is certainly being innovative. Enough to keep my interest in their latest offerings. But IOS is too Leapfrogish for me and I also have zero interest in slipping into the Apple Desktop ecosystem from my Windows PC. Compared to Firefox/Chrome Safari is crap and the constant Quicktime nags are annoying. I test drove an iPhone 6 for two weeks to objectively evaluate the option. Pass.
  • melgross - Monday, September 28, 2015 - link

    Objectively? Is that even possible? Nope!
  • Featherinmycap - Monday, September 28, 2015 - link

    What is a Quicktime nag?
  • Shiitaki - Thursday, October 1, 2015 - link

    Quicktime Nags? Please inform me! What have I been missing with my house full of Macs? BTW, Chrome and Firefox run great on Mac. Just out of curiosity, how well does the phone you current have integrate with your Windows PC? Do you even understand why I ask that? Do you wonder what you are missing?
  • Fidelator - Monday, September 28, 2015 - link

    All fine and dandy but the performance increase should have been taken advantage of by upgrading to a 1080p display, there's visible pixelation on rounded objects on the 6, even at a reasonable distance, its 2015, a 720p screen is an insult for the price prople pay for the 6S, this is unacceptable, the 6S+ seems like the only good option if you want an iPhone, also considering the lack of OIS on the regular 6S this statement is solidified.

    Before the fps gains argument, this is currently irrelevant as no mobile game will push this GPU to it's limits, even at 1080p, given the case, Apple might as well start shipping the 6S with a 854x480 screen and calling it a day over the superior gaming performance.

    Hopefully the reviews will be separated, as the 6S+ is a whole other device, and a far more acceptable one while at that, the 6S is a shame.
  • blackcrayon - Monday, September 28, 2015 - link

    Nearly invisible pixelation vs. highly visible lag from your favorite 1440p vendors? I'll take the former.
  • tyj - Monday, September 28, 2015 - link

    I've always known Nexus 6 storage perf was bad, due to encryption by default and due to Android not being able to use the dedicated encryption chip yet. Out of curiosity, were these benchmarks done an an encrypted (stock) Nexus 6 or an unencrypted unit?

    It'll be interesting to see how the Huawei Nexus 6P rates as well - these benchmarks may be enough to sway me to the iPhone this year. Tough decisions...
  • tipoo - Monday, September 28, 2015 - link

    I read that it was because the hardware encryption on that chip was slower than software, somehow. Not because Android just can't use it yet.
  • name99 - Monday, September 28, 2015 - link

    "Twister", huh?
    I thought this would be a three-cycle micro-architecture going eastward (cyclone, typhoon, hurricane) but they seem to have paused over the Continental US. This in turn suggests that next year's model will be like the three before it, only better in all dimensions, and it will be at least 2017 before we see a truly new micro-architecture.

    The most interesting fact about Twister itself seems to be that this was the CPU for which they decided to improve the uncore. Look here (comparing iPhone6S to the best retina Macbook score on the site)
    https://browser.primatelabs.com/geekbench3/compare...
    The point of interest (IMHO) is how close to identical all the memory scores are, and likewise the memory dependent benchmarks (things like the FFTs and blur/sharpen).
    Time was (like as recently as a month ago) that people were saying Intel's great strength was their uncore --- other people might put together a great core, but only Intel had the expertise in the memory controller and the cache management algorithms to get the data flowing optimally all the way from RAM to registers. Not so much anymore...

    The one place Intel probably is still substantially ahead of Apple is in tech related to coherence and consistency for many cores, the sort of stuff needed to make an 18 core Xeon work well. I'm guessing Apple won't have much incentive to push in that particular direction until they ship their SoC for the Mac Pro...

    One final point. I'm not convinced that 3Dmark Physics is purely a test of frequency as you suggest; I think there is more going on here. Compare, for example, the Surface Pro 3 to Samsung and iPhone 6S numbers.
    My current hypothesis is that this matches something else we have observed, that Apple SoC seem to have high L2 and L3 latencies. It is possible to design a cache that is segmented into differently behaving portions. A simple example would be a follows: Suppose your cache has 8 ways. You put two ways in the A (fast) sector and 6 in the B (slow) sector. Every L1 miss FIRST probes the A sector and only on an A miss probes the B sector (which is a drowsy cache). A hits are fast, B hits require at least one more cycle to probe the extra ways, then an additional cycle or two to wake up the drowsy line (and that line is then swapped with an A sector line).
    Cache designs like this have to been modeled to save 40 to 80% power, at the cost of 2 to 5% performance, which seems like a good tradeoff. (I don't know if anyone has implemented one, but Intel implemented something like this for the Ivy Bridge L3, where the B partition was, I think, not drowsy but simply shut down.)
    In essence this is an implementation of the L1 idea of way prediction applied to L2 or L3 in a way that makes sense in those contexts.
    The one place they expose a glass jaw is in code that more or less randomly bounces around the address space --- stuff like L2 latency testers and (as far as I can tell) 3dmark Physics --- where the heuristics for what to put in the A vs B partition and when to swap them fail.

    There are ways to make this more sophisticated. You can dynamically change the size of the partitions so that if your entire L2 or L3 is being used aggressively you make everything A sector. (This is essentially what Ivy Bridge does, swapping between use of the entire L3 and use of a smaller A partition dynamically). You can have the A partition be one way in size, or two ways, or four ways, or the full eight ways. I don't think Apple is dynamically sizing the cache yet --- if they were, then the latency and Physics numbers would show the latency for just the A partition and be somewhat improved.
    You can also add a third state of dead-line prediction and (to save power further) just drop dead lines in the B-partition. It's possible that Apple is already doing this; if not it's one more thing that can be added for the future.
    There are also other tricks available to make L2 that much more performant. One that is simple enough that I am guessing Intel and Apple already use is to mark I-lines in the L2 as higher priority than D-lines because the CPU hurts more when it misses in L1-I than when it misses in L1-D. The standard algorithm for this is called Code Line Preservation (CLIP).

    To summarize, I think Apple have made the right choice with the cache here. It obviously can be shown to be slower than a traditional cache if tested in exactly the right way, but for "normal" code it seems to be every bit as fast as the simulations show it should be, while reducing power substantially, which then allows more power for things like branch prediction and 6-wide OoO.
  • tipoo - Monday, September 28, 2015 - link

    Yeah, I thought the core count/frequency explanation for the physics scores never really made sense. The A8X with three cores didn't improve it as much as you would think it would given 50% more cores, whereas Android phones seemed to happily scale on it. So there must be something else going on in the SoC design. Interesting theory about caches.
  • WoodyPWX - Monday, September 28, 2015 - link

    Any chance you can compare iP6s+ with Note 5?
  • Shalmanese - Monday, September 28, 2015 - link

    So one of the big arguments people make towards moving to 2Gb of RAM is so that Safari pages don't refresh from the web when you go back to a previous tab. But desktop Safari users never encounter this problem, no matter how little RAM they have, because Safari caches tabs to disk when it runs out of memory.
    I get why the iPhone 2G decided not to implement swap files with it's abysmal I/O performance and limited write cycles. But this latest generation of phones is apparently using something close to a desktop SSD solution and yet iOS still has hard Out of Memory errors rather than introducing swap. Is there a reason at this point why iOS doesn't just mimic desktop OS behavior with regards to memory?
  • amonduin - Monday, September 28, 2015 - link

    I think there are probably two reasons for this.
    1. Even though storage IO has gotten much quicker it is still quite a lot slower than lpddr4 memory access and so allowing apps to use swap memory for their data would likely (in Apple's mind) lead to a degraded user experience.
    2. Limited storage. Many Apple products are still limited to only 16 GB of internal storage and filling it with swapped memory data could fill it up rather quickly. This second issue is one which they should have alleviated by now with larger base storage classes but apparently the money is too good to pass up.
  • name99 - Monday, September 28, 2015 - link

    iOS provides a paging infrastructure. Code is paged in, and memory mapped files are supported (and encouraged by Apple as the best way to access files).
    What is NOT supported is demand paged writes. I suspect this is because of either the power cost (flash writes are expensive in power) or the jerkiness this might add to the UI.
    This seems to be the common consensus for mobile OS's. Android and Win Mobile make exactly the same policy choices.

    Of course if Apple can switch to some alternative to Flash in future iOS devices, the tradeoffs will differ, and at that point swapping might become a sensible choice.
  • tuxRoller - Monday, September 28, 2015 - link

    And this is why all the other phone oems area behind Apple. They are proactive in their improvement attempts instead of adding halfbaked software and calling it a day. Samsung has the expertise to provide the fastest, bar none, phone storage solution, but no, better to be conservative and copy Apple.
    Apple, first to: high dpi screens, desktop class single core performance, reliable fingerprint scanner (speaking of, where the hell are those amazing Qualcomm fingerprint scanner that can fit on the screen?), desktop class storage performance.

    I'm still not going to buy an Apple product but this is getting to be ridiculous. With all of the various Android manufacturers you'd think at least one would produce something as good as Apple's hardware.
  • tempestglen - Monday, September 28, 2015 - link

    please run SPEC06 because A9 is a desktop class cpu and with 2G ram.
  • tuxRoller - Tuesday, September 29, 2015 - link

    Sure. Get me a device with an a9 soc and a copy of spec06 that will run on it.
    I'll give you my address later.
  • Buk Lau - Tuesday, September 29, 2015 - link

    apple questionably introduced high dpi screens to desktop, but they were years behind in terms of resolution in mobile, so I don't get where you got that from
  • Argosy - Tuesday, September 29, 2015 - link

    Buk Lau,
    Really? There was a higher than 300 dpi display on a mainstream mobile device before the iPhone 4?

    Apple was late to bigger than 4" screens, but they were at least on the leading edge (if not the leading edge) of high dpi displays.
  • Blark64 - Monday, October 5, 2015 - link

    They were the first with a 300+ DPI screen on mobile, in 2010. See:

    http://www.engadget.com/2010/06/22/iphone-4-review...
  • KoolAidMan1 - Tuesday, September 29, 2015 - link

    But Apple just sells mid range hardware at a premium price!!!!

    /s

    Doesn't matter, Dailytech style idiots are going to keep imagining that core counts or useless features like air scrolling somehow matter.
  • Kuzi - Tuesday, September 29, 2015 - link

    iJoshua Ho and iAnandtech (Apple owned) gang know very well that all Galaxy phones come with their own stock browser that is much faster than Chrome, yet they don't use that for their biased testing.

    Below are the numbers I get for the Galaxy S6 stock browser:
    Kraken 1.1: 3898ms (59% faster than iAnands Galaxy S6 numbers)
    Google Octane v2: 9501 (45% faster than iAnands Galaxy S6 numbers)

    3dmark 1.2 unlimited:
    Score: 25693 (14% faster than iAnands numbers)
    Graphics: 28775
    Physics: 19785

    Regarding the 3dmark Physics test and iJoshua's iBullshit comment, the low number for iPhone is because of the dual core CPU. Android OS utilizes all 4 cores very well and it shows here. Even the iPad Air 2 only gains a bit from the triple cores. Seems like Apple has some work to do on multicore optimization on iOS.

    Finally, other reviews have found the iPhone 6S battery does last slightly less too, so there is no need to sugar coat it. Don't worry iJoshua your Apple lords won't get angry if you admit there is a 5% degradation compared to last generation.
  • blackcrayon - Tuesday, September 29, 2015 - link

    There isn't enough information in the benchmarks here to make a determination on how well either OS uses multiple cores. But it's pretty clear (not that we even need benchmarks for it) that having a smaller number of fast cores is more valuable to a mobile device (and most consumer computing devices in general, PCs, etc) than a large number of slower ones.
  • Kuzi - Tuesday, September 29, 2015 - link

    You can check the link below for Android multicore analysis by iAndrei:

    http://www.anandtech.com/show/9518/the-mobile-cpu-...
  • redvodka - Wednesday, September 30, 2015 - link

    The common browser that can be used by all Android phones is Chrome so how can you assume the reviewer is biased? But sure even if you do use the Samsung optimised browser results, it is still not the clear winner.
  • Kuzi - Friday, October 2, 2015 - link

    Samsung's Worldwide Android market share is probably 50% (I don't have exact numbers), so that's a big chunk of the phones on the market that include the faster browser. Also HTC, Sony, Asus, Huawei and probably others include their own stock browsers. So that's most of the Android phone markers that include an optimized stock browser that is generally faster than Chrome.

    Safari is the optimized browser for iPhones/iPads so I only see it fair to use the optimized browsers of other phone makers when benchmarking.

    And I do agree that the IPC is noticeably faster on A9 SOC than any other available mobile SOC, and graphics is even better and is a great achievement.
  • MrSpadge - Tuesday, September 29, 2015 - link

    It's a shame they don't put Win Phone onto this hardware.
  • AnnonymousCoward - Tuesday, September 29, 2015 - link

    When you say "4K video", you've gotta include the framerate. It's not a given.
  • Tech_guy - Tuesday, September 29, 2015 - link

    what's more impressive vs Note 5 is this
    Note 5 3D performance is weak especially having to push the extra pixels on top of Mali T760 MP8 being about 200% weaker than A9 has. I love gaming on my phone so Note 5 is horrible in this area, unless the developer renders their games at 720p rather than Quad HD. Note 5 CAN'T handle in 3D demanding tasks at it's native resolution.

    GFXManhattan onscreen (device's native resolution)
    iphone 6s 55.9 fps
    iphone 6s + 38.4 fps
    Note 5 14.6 fps

    GFXBench Manhattan offscreen 1080p
    iphone 6s 39.4 fps
    iphone 6s + 39.4 fps
    Note 5 19.9 fps
  • AnnonymousCoward - Tuesday, September 29, 2015 - link

    No matter what Apple does, the phone still doesn't have a back button and you still can't install a different default browser (a better one like Firefox with AdBlock with the topbar always visible).
  • redvodka - Wednesday, September 30, 2015 - link

    FYI: The new iOS9 has a back button now when you open new apps from the previous app.
    True, you cannot install a default browser but you can still install Firefox and when clicking on links, you can select' Open in Firefox" .

    Also the new Safari in iOS9 allows the user now to install adblockers.
  • AnnonymousCoward - Thursday, October 1, 2015 - link

    Except Firefox isn't even available for iOS.

    And that back button implementation sounds a lot less flexible and global that the one on my Android phone's chassis that works on everything.

    And is it possible to push 1 button to exit sleep and resume operation, or do they still force the user to slide their finger across the screen?
  • Argosy - Thursday, October 1, 2015 - link

    Not sure a software back button sounds less flexible. In iOS9 the "back button" is labelled to show where you are going back to.

    Three generations ago (iPhone 5s) you have been able to wake a locked phone from sleep with home/TouchID.
  • AnnonymousCoward - Friday, October 2, 2015 - link

    redvodka said there's a back button _when you open new apps from the previous app_. That's a specific case, so that's why I said it sounds less flexible. It also takes up screen space compared to a permanent button on the chassis.

    Regarding the 1-button to exit sleep, I was talking about when there's no security set, and you use the power button to exit sleep. This is especially annoying on older iPads that don't have a fingerprint scanner to begin with. You can't just hit 1 button to resume operation; they make you slide a finger across the screen. That's a waste of users' time, it's difficult for elderly or handicapped people, and it's a problem if an iPad is mounted in a special way for a unique application. Apple is too incompetant to understand that.
  • Blark64 - Monday, October 5, 2015 - link

    In use, there is no issue with the lack of a back button. In situations where one is necessary, it is there on the screen. In situations where a back button is not necessary or useful, it isn't there. There is one obvious hardware button, the home button, and that has the same function everywhere: to take you back to the home screen.

    And very few I have ever seen uses the power button to wake a device from sleep: they usually use the home button (which doubles as the touchID sensor on newer phones and iPads). Whether or not they must swipe depends upon how recently the iPad entered sleep. Swiping to unlock is a necessary trade off for a mobile tablet, since it could otherwise be woken, and have its powered drained, by unintended button presses in a bag or purse (all smartphones and tablets have a similar system by default). I have never heard anyone complain about the "waste of time" that entails, and it seems you are really reaching for something to complain about.
  • Macman2288 - Tuesday, October 20, 2015 - link

    Well said, there are plenty of new advantages to android worth mentioning, makes little sense to dig up 2012 features that iOS clear has addressed.
  • Harry_Wild - Tuesday, September 29, 2015 - link

    Very interest! the new A9 SoC just so fast! Almost twice as fast as the A8 in some cases! I guess I have to think about updating my 6 to either 6S or 6S+! Not sure about the "Rose Gold" color however! LOL!
  • Bfree4me - Tuesday, September 29, 2015 - link

    I still believe that we are now at a point of diminishing returns on mobile phones. Look no further than Ebay to see what a 6 month old phone is fetching. A mid to top tier device has a residual value of 60% or less.

    Apple, Samsung and others have to continously keep coming up with incremental features to sell these kits. Soon, most consumers will wake up and smell the roses like most have done on desktops . Each and Every mobile device today has a good camera, screen and battery. There really is no point in paying for a new device each and every year. The only innovation that is truly lacking is battery power. If a modern mobile device launched today and it had 48hr or more batter life it would sell like Hotcakes.
  • ws3 - Tuesday, September 29, 2015 - link

    I don't see the appeal of a 48 hour battery. I charge my phone every night while I'm sleeping, so if it gets 16-17 hours, I'm good. I'm pretty sure that sleeping is a common daily activity for smartphone owners so enough battery to last for the full waking period seems to be entirely sufficient.
  • blackcrayon - Wednesday, September 30, 2015 - link

    I can see a use for, every once in a while, being able to withstand heavy use in one day without charging. But that rare event is easily covered by a battery case, and keeps the phone thinner and lighter the other 95% of the time.
  • Argosy - Thursday, October 1, 2015 - link

    If you conducted a poll, I suspect most people would feel camera and battery needs improvement. It is just this year where more than a few phones are being touted by reviewers as being as good or better than the iPhone camera. How many people think their phone camera take good photos indoors/low light? So I think battery and cameras both have a ways to go.

    People were saying there is "no point" in better desk top processors for 10 years. We are finally there for a majority of desk top users. It has almost always been the case where you build a better processor and people will find a compelling use.
  • byrons - Tuesday, September 29, 2015 - link

    One thing that hasn't been measured by other reviewers is the impact of the new always-on "Hey Siri" feature. That's going to affect the battery comparisons vs. the 6 even while idle. Given that Low Power Mode disables Hey Siri on the 6s, it's likely non-trivial.

    So if you have a chance, please consider a comparison with and without this feature enabled.
  • tipoo - Thursday, October 1, 2015 - link

    I don't imagine it will be too big, since they're using an on-SoC solution like Motorola did back with the first X. Low power mode disables it, but I think that's more "every bit counts" thinking to get that extra bit. On it's own without all the other things low power mode does, always on Siri probably won't do much.
  • dean85 - Wednesday, October 7, 2015 - link

    We provide convenient, cost-effective way to purchase your cell

    phone. you email us and obtain information about the latest

    products at a very competitive price.

    Our latest promo is BUY 2 GET 1 FREE!!!

    INTERESTED BUYER SHOULD CONTACT US THROUGH
    CONTACT EMAIL: shophere.inc@gmail.com

    Apple iphone 6s unlocked for $600usd

    4G LTE: all models
    Sim: Nano-SIM
    Announcement: Sept 9th
    Release: Sept 19th
    Dimensions: 137.5 x 67 x 7 mm (5.41 x 2.64 x 0.28 in)
    Weight: 113 g (3.99 oz)
    TouchID: 500 dpi pixel density fingerprint sensor (Touch ID on all models)
    Display: LED-backlit IPS LCD, capacitive touchscreen, 16M colors
    Size: 960 x 1704 pixels, 4.70 inches / 5.5 inches (~416 ppi pixel density)
    Multitouch: Yes
    Protection: Shatter proof sapphire crystal glass, oleophobic coating – water

    resistant
    Sound: Alert types Vibration, proprietary ringtones
    Loudspeaker: Yes
    3.5mm jack: Yes
    Memory Card slot: No
    Internal: 16/32/64/128 GB, 2 GB RAM
    Data GPRS: Yes
    EDGE: Yes
    Speed: DC-HSDPA, 42 Mbps; HSDPA, 21 Mbps; HSUPA, 5.76 Mbps; LTE, Cat6, 300 Mbps

    DL; EV-DO Rev. A, up to 3.1 Mbps
    WLAN: Wi-Fi 802.11 a/b/g/n/ac, dual-band, Wi-Fi hotspot
    Bluetooth: v4.0, A2DP
    NFC: Yes
    USB: v2.0
    Camera: Primary 8 MP, 3264 x 2448 pixels, autofocus, dual-LED (dual tone) flash.

    Features Simultaneous video and image recording, touch focus, geo-tagging, face

    detection, HDR panorama, HDR photo
    Video: 1080p@60fps
    Secondary: Yes
    OS: iOS 8
    Chipset: Apple A8
    CPU: Dual-core 2 GHz
    Sensors: Accelerometer, gyro, proximity, compass
    Messaging: iMessage, SMS (threaded view), MMS, Email, Push Email
    Browser: HTML (Safari)
    Radio: No
    GPS: Yes, with A-GPS, GLONASS
    Java: No
    Colors: Space Gray, White/Silver, Gold
    - Active noise cancellation with dedicated mic
    - AirDrop file sharing
    - Siri natural language commands and dictation
    - iCloud cloud service
    - iCloud Keychain
    - Twitter and Facebook integration
    - TV-out
    - Maps
    - iBooks PDF reader
    - Audio/video player/editor
    - Organizer
    - Document viewer/editor
    - Photo viewer/editor
    - Voice memo/dial/command
    - Predictive text input
    Battery: Non-removable Li-Po 1,810mAh battery (6.91 Wh)

    INTERESTED BUYER SHOULD CONTACT US THROUGH
    CONTACT EMAIL: shophere.inc@gmail.com
  • mkimid - Thursday, October 8, 2015 - link

    I hope we can see the real world battery life test when some higher load on CPU/GPU; such as GFX benchmark, or 4K video recording and playing, etcs. if the battery life is affected with CPU usage(even it is not 100%), if more APU parts are running together (GPU); I feel that it will be more serious difference of results.
  • twin-pt - Friday, October 16, 2015 - link

    When can we expect your full review? I'm thinking on buying an iPhone 6s Plus 64GB...
  • Macman2288 - Tuesday, October 20, 2015 - link

    They really are taking forever to post the full review. I am not sure why, but last year it took them only a few weeks to complete the full review. Either they found something strange or they are understaffed
  • twin-pt - Sunday, October 25, 2015 - link

    Yes, it's probably both... Let's wait a bit more... I really want one 6s Plus but anyway it's out of stock in Europe (at least in Portugal)...
  • ASEdouardD - Monday, October 26, 2015 - link

    Do you plan on releasing a complete 6s review this century or the next?
  • a619 - Monday, October 26, 2015 - link

    For a month now, I've been visiting this page every morning in hope of finding the reviews. :(
  • Ishwa - Monday, October 26, 2015 - link

    Same here - Checkin' every day only to be disappointed.
  • Lavkesh - Tuesday, October 27, 2015 - link

    This is getting ridiculous now. A month on and still no review?

Log in

Don't have an account? Sign up now