Hacker News new | past | comments | ask | show | jobs | submit login
iPhone 6 and iPhone 6 Plus: Preliminary Benchmark Results (anandtech.com)
85 points by srikar on Sept 22, 2014 | hide | past | favorite | 105 comments



The A8 chip is "only" a dual-core and clocked at a relatively modest 1.4GHz. It's heartening to see it beat competing 2+GHz Quad-cores. To me, it looks like Apple is comfortable enough with their marketing strategy to stay away from the pointless GHz rat-race that exists in the Android world.

http://en.wikipedia.org/wiki/Megahertz_myth


The ONLY CPU benchmark that was not browser-based (and therefore running wildly different software) is the 3DMark Physics test, and A8 lost bad on that test.

Everything else was either single threaded (all the browser benchmarks), in which case dual vs. quad is irrelevant, or they were GPU bottlenecked in which case you aren't talking 1.4GHz dual core, and on the GPU front Apple spends a huge amount of transistors on that. They throw oodles of power at the GPU.

Also thanks to power management the vast, vast majority of those "2+GHz Quad-cores" were not running at 2+GHz, they were running at closer to 1.4-1.6 GHz. Apple is really good about settings its max frequency to something sustainable, whereas everyone else advertises burst frequencies.


Could you pls explore browser benchmarks being single threaded? I mean, when I use Chrome or Firefox on Android isn't the browsing experience implemented as multi-threaded (e.g. when I have the 8-core S5)?


There are multiple threads, but the benchmarks are specifically looking at JavaScript performance (as it's the only thing they can actually measure - anything claiming to measure drawing performance is a lie, whether or not they realize it, but I digress), and JavaScript is all single-threaded.

There is going to be painting and such while the benchmark runs, which will use a second thread, but that's about it. And that's going to be relatively light compared to actual usage where you're scrolling around and such.

The majority of the other threads of a browser are I/O bound (networking, disk, etc...) and won't be using much of the CPU.


It's a false claim that there is GHz rat-race in the Android world. There is one Android SoC maker of note, and that is Qualcomm, who achieved total dominance of high end smartphones by anticompetitive behaviour.

Just because Qualcomm's design achieves high clock speeds doesn't mean there's a rat-race in the Android world. Since Qualcomm dominates, there is no race at all. Second, Qualcomm's competitors are much more moderately clocked. Samsung's, Intel's and Nvidia's current high end chips don't even touch 2 GHz yet.

For a rat race you need rats and a race. But there's only one rat and no race.


What was Qualcomm's anti-competitive behavior?


Their SoCs were too good.

Critically even though they may not always have the best CPU cores or best GPUs, they do have the best radios. And integrating that into the SoC is just better from a power perspective, making Snapdragon overall the best SoC package for a phone.


I don't see how that's anti-competitive.

Isn't that just... competitive?


Yup! I have no idea what the other person was talking about tbh.


I accept your point.


Apple is better on Javascript benchmarks. If you look at the 3DMark physics engine the iphones are towards the bottom of the list as you would expect. This tells me that the javascript engine for ios / safari is far more optimized than the ones for Chrome / Android.

In day to day life your performance is going to be in between these two extremes.


I can't attest the value of this test https://www.youtube.com/watch?v=3P9vDAY0U0A

They show how a Moto E is more responsive than a Galaxy S5. Full stack system performance, when doable, gives you this kind of leverage.


I carry a Moto E daily. Out of the box it is indeed reasonably snappy, but as soon as you have a few apps running it really starts chugging. Great device for the target user, but if you actually load a bunch of apps then you start seeing the limits of it.

Mind you I'm not a game player or anything, this is just the normal stuff of having a podcast app, spotify, HN and a few others daily used apps. Great phone, but it can't compare with a flagship. (I carried a Nexus 5 before)


Thanks for the info. So I guess high end devices may lag but it's a fixed overhead.


Methinks Apple has decided/realized that specs are a cop-out. Customers are, all too often, inundated with numbers not so much to prove how good the product is, but to conceal that it's not good enough. "Meh, the new iPhone has a mere 1.4GHz dual-core, but our product has a 2+GHz quad-core!" really means "here's some numbers that look overwhelmingly superior, in hopes that will distract you from the fact that it isn't."

Apple went "retina" to end the pixel-count competition: if you can't see the pixels, you don't care how many there are. Apple refused to state RAM size for mobile devices from the beginning: user doesn't have an option, and rarely (if ever) notices the limit, so the actual memory size doesn't matter. Processor count & speed is irrelevant if it's used efficiently enough to practically outrun the competition.

Sure, Apple does provide some specs. People are too used to seeing such numbers, despite few needing to see them IF the implementation is done right. People want options, so some specs are flexible (at a price). But remember that behind the scenes at retail, the devices options are labeled little more than "Good, Better, Best": most users don't want better numbers, they want superior performance with whatever it's got. Those gloating about size are likely less able to use it.


This is something I've been harping about for a long time. Apple is a _user experience_ company, not a hardware or software company (perhaps they're the only UX company out there).

There's an old saw about the railroads getting killed by the tractor trailer in the 20th century because they thought they were in the railroad business, when in fact they were in the transportation business. This time, Apple knows exactly what business it's in, while the competition seems to have no idea (compare the highly-curated experience of shopping/buying/shipping/using an Apple product to the Android or Lenovo analogue).


Apples doesn't advertise clock rate or RAM size. You can't even find it on the tech specs pages. The most they do is tell you the chip's name.


Apple introduced 'retina hd' this year; they are still doing 'pixel counts' by other means


There's some documentation of the A7 (the previous chip) here: http://www.anandtech.com/show/7910/apples-cyclone-microarchi...


I can name many friends of mine that talk down towards the iPhone because their <current state of the art android phone> has more cores or ram or ghz, but what does that actually materialize into? If the extra performance doesn't actually do anything for the user then why waste the money on it? Apple has an amazing UX and doesn't need to throw pointless numbers at you to get you to buy their phone.


The really interesting data point to me is the 6+'s 13.7 hours of WiFi browsing versus the LG G3's 8.8 hours. Both are 5.5" screens with ~3000 mAh batteries, but the iOS/A8 combo lasts more than 50% longer.


The LG G3 has a 1440x2560 QHD screen. It takes more power to dive more pixels.


So Android pushed 77% more pixels using only 50% more power? Sounds like Android is more efficient!

Obviously power is more complex than just that, but seriously Android's reputation for being inefficient is really not justified due to things like this, where resolution differences are completely ignored.


A minor correction: The iPhone 6+ renders at 1242 x 2208 but scales it down to 1920 x 1080. So 30% more pixels at 50% more power?


It's the physical pixels that suck most of the power, so "77% more pixels" is more accurate.

Certainly the GPU on the iPhone 6+ is going to take more power because it's rendering at 1242x2208 rather than 1920x1080, but that's a minor effect compared to the physical screen power draw.


That presupposes that lighting up and drawing pixels is most if the overall power budget. Moreover, the biggest component of screen power usage is the backlight, which doesn't scale linearly with the number of pixels.


> That presupposes that lighting up and drawing pixels is most if the overall power budget.

That very much IS most of the overall power budget. That was very not a guess on my part.


See: http://www.displaymate.com/Smartphone_ShootOut_3.htm. At the same brightness level (Anandtech uses 150 or 200 nits I think), the FHD displays use only 15% more power than the iPhone display despite almost 3x the pixels. And total display power is probably about half the power budget.


Read how that test is run. It's showing a static image. There's nothing changing, there is no rendering in that test at all. So of course resolution didn't have an impact, the test is built to test purely the display's power draw and does excactly that.


It'll be interesting to see how much Android improves with Android L; an early June build gave the Nexus 5 around 30-40% extra battery life. I believe that sound latency is also reduced by an order of magnitude in L, and there are many big improvements to the camera API. Interesting times indeed!


It'll be interesting to see how that changes with Android L. My Nexus 5 gets massively better battery life (~30% for me) on the developer preview.


This is what happens when you control both hardware and software.


No, this is what happens when your screen is less resolution :)


are you saying that Apple made the right tradeoff?


No?

Someone made a claim that controlling the hardware and software stack was the driver for being able to do well, powerwise.

I'm saying that controlling the hardware and software stack probably had less to do with this performance than the GPU only needing to push 40% as many pixels. I would also bet the overall hardware PM is better in Apple's chips.

Software wise, there is only so much you can do that has any real impact, given that screen is 40% of power and cell is another 20%.

Past that, only real way to save power is often to make stuff run faster.

But to give you some perspective, if you take any of these phones, turn off cell usage and shove them in a drawer, they should last at least a week.


but given all that do you think apple made the right tradeoff? between ppi and battery that is.


Maybe? The truth is I feel like it didn't matter what they chose.

Apple is very good at convincing people of what they want (i don't think they simply give people what people already wanted). In a case like this, IMHO, they could have had slightly better battery life than the 5S, and said "we've packed a lot more power in the same battery life", and people would have likely been fine with it.


To give you a faint idea of how meaningless battery benchmarks are vis-à-vis real life usage, here are results from Phone Arena - http://www.phonearena.com/news/All-bow-to-the-new-endurance-... - 9+ hrs for Xperia Z3 and 6+ for iPhone 6+ which is lesser than the S5.

"We measure battery life by running a custom web-script, designed to replicate the power consumption of typical real-life usage"

In other words I would be very surprised if any of these benchmarks translate to anything close to the numbers they proclaim.


Showing two different benchmarks doesn't actually prove that battery benchmarks are meaningless.


Well they are both web browsing battery life benchmarks. If loading different web pages shows 5+ hours of discrepancy then yes from a user standpoint they are precisely meaningless.


No, it might mean that one of the benchmarks is meaningful and the other is not. Or it might be that they test different scenarios and could be useful to different groups of users.

What the discrepancy tells you is that there's something odd about the benchmarks that could stand to be better investigated/explained.


> there's something odd about the benchmarks

You say odd, I say meaningless :)


Ok, but they're not the same thing. Odd might just mean a different set of underlying assumptions that cause the numbers to misalign, or perhaps a bug.


I was waiting for some independent battery tests and this kind of confirms what I suspected would happen. Apple is back to absolutely destroying Android on the battery front.

Some might not remember, but the drive to bigger Android screens was actually mostly driven by battery life. Android has always been subpar in this critical respect, the first devices were absolutely terrible, I suspect primarily because Dalvik is nowhere near as efficient as cross-compiled objective C and the rather more restrictive background task management on iOS.

iPhone 5's were holding their own and then some against the much larger battery Android devices, so now that they batteries are the same on the 6's there is no contest. Yes, bigger screens do eat more battery, but the ratio of battery/screen size actually favors bigger screens, ergo why tablets get incredible battery life. (and why the Note has always been a great battery performer)

I've carried a Nexus since the very first one, but I think I might jump to a 6+, the advantages are too much to ignore. I'll wait to see what the Nexus X is first as I really love Android these days, but I sure would love that battery life and camera.

Nice to see Apple put the screws down.


Yeah, this would be all very well and good if Apple started shipping phones with larger batteries. The fact that battery life per mAh (may be) better on iOS than Android means nothing if Apple keeps on rejecting larger batteries in favour of thinness.

I know a lot of people would much prefer a slightly thicker iPhone 6 (that didn't require the camera to protrude, for one) if it meant a larger battery.


I'm surprised no one mentioned the ability to use custom ROMs and kernels on android devices with optimized CPU governers and overclocking on demand and how that affects the phone's performance.

Android stock branded phones are packed with bloatware from the carriers. This is something apple actually has a grip on due to how closed their platform is.

I'm running c-rom with lean kernel on my Galaxy Note 3 and running just one of those benchmarking programs, I get results that are a lot better than what anandtech shows. Still not better than apple, but pretty close to them for a last generation phone.

I'll try to run all of the benchmarks sometimes this weekeend and do a post.


Could you please elaborate what is "c-rom with lean kernel" and how can I run it? (on the original HTC mini). I bought it because I wanted a non-Android smartphone that's not huge and decently built, but I'm tired of HTC Sense.


(S)he's talking about running a custom operating system on their phone. Check out xda-developers if you want a overview on the topic. I haven't actually done it myself, but they're the people to talk to.

Side note, I'm confused, you bought an Android phone because you wanted a non-Android smart phone? HTC Sense is still Android, just with a custom skin.


Sorry I can see how this is confusing, but it's basically what mpthrapp said.

I've got a unbranded version of android suited for my phone with an optimized kernel (lean kernel).

It runs a lot faster, the battery lasts longer, and I don't have to deal with 200+ bloatware apps on the phone...

Here's the page for C-ROM. See if it's compatible with your device.

http://www.c-rom.org/

lean kernel for note 3 http://forum.xda-developers.com/showthread.php?t=2488233

Your best bet is to just get on xda-developers forums and go to your device's section.


oops, mis-edit. I wanted a non-huge Android phone.


I'd like to see some PC-phone comparisons now that they're starting to approach parity. By my calculations, the new iPhone seems to be roughly as fast as a 3 - 4 year old Macbook Air.


There's a handy Reddit thread[0] with reports of actual "real world" usage.

[0]http://www.reddit.com/r/apple/comments/2gzxx1/post_your_batt...


I don’t think that’s very valuable. We are (sadly) at a point where you can ruin your battery life in iOS depending on how you use it. With great power (to actually run things in the background for real) comes great responsibility (to micromanage your apps) …

The improvements we see in benchmarks can thus easily be lost in the noise.


I'm always confused by the terms "Usage" and "Standby". Between those two which is "Time off charger" and which one is "Screen Time"?


I think there's a pretty good explanation here: http://www.scottyloveless.com/ios-battery-life/


Ok that explains it a little bit.

So Standby time is basically "Time off Charger", and Usage is supposed to be "Screen on time"? Except I have a feeling that "Usage" includes any time when your phone is not sleeping (which would include not just the time when you have the screen on but also any time when the processor is doing stuff in the background).

What draws me to that conclusion is the ratio of usage to standby time that some people in that thread have. I have a hard time believing that most people are staring at their screens for half of their waking life. I mean even if you didn't leave the house all day you'd still have to take your eyes off the thing to prepare food, urinate, put your clothes on.


Yep, listening to podcasts (with the screen off), for example, is definitely “Usage”.

I’m not quite sure how the phone counts background activity that doesn’t notify the user or require user interaction (e.g. Mail checking for new emails, …).


I think that's exactly right.


Wish they would include the iPhone 5 results. Most people keep a phone for two years.


You can run the benchmarks yourself!

http://www.webkit.org/perf/sunspider/sunspider.html


It's not convenient, but you can always go back to their 5s review to see the 5/5s difference.

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


Yeah. And 4 and 4s. I'm interested in what upgrading will give me, not really how it compares to other high end phones (not switching from iOS). I doubt many 5S users are upgrading to 6, but a lot of 4S and 5-users.


I think the real standout here is NVIDIA. Shield was such an interesting product from the very beginning, it's nice to see how it is actually holding up against some of the "big boys".


To be fair, the Nvidia shield mentioned in these benchmarks is a 8" tablet, not a smartphone. I think k1 hasn't yet been shown in a smartphone form factor. Given the limitations caused by size, it's not really a fair comparison.


Especially considering that these devices are all thermally limited: http://cdn.arstechnica.net/wp-content/uploads/2013/11/SoC-th...


It seems odd they report 10 hours web browser battery life for the iPhone 5. I'd be lucky to get 1-2 hours of constant browsing.


I suspect you're eligible for a brand new iPhone 5.

Check the following: https://www.apple.com/support/iphone5-battery/


Your battery us probably dead. Maybe get that checked?

For what it’s worth, my 5s gets about six hours of intense use. 50% brightness, Twitter, Safari, RSS reader, mix of WiFi and cell towers, mostly good reception. Five hours with only cell towers, average to good reception.

However, since the iPhones have such tiny batteries their battery use is also highly dependent on how you use your device. With something downloading (especially over LTE) I can practically see my battery draining and it gets quite hot.

Bigger devices with much larger batteries are a bit more sedate in their reactiveness to that, e.g. my iPad is much more constant in its battery usage. No matter what I do, it always gets nearly the same battery time. (Maybe because battery usage is much more dominated by the screen which has a constant battery use, so the other two big battery consuming components which are variable, radios and SoC, have a smaller difference in impact between being nearly idle and running at full tilt.)


They are using Wifi and pausing between page loads (so that fast phones don't load 4x as many pages and lose the test).

On Wifi 10 hours sounds closer to my experience than 1-2 hours.


My phone is 2.5 years old at this point. Maybe I've been losing battery capacity so gradually I've never noticed.

That would be amazing to have 13 hours of browsing with a new phone. (as reported in the article)


The iPhone 5 was only released 2 years ago (21 Sept 2012), yours can't possibly be 2.5 years old.


Ok, apparently I have a 4S. Maybe it's older than I thought.


I know someone with a 4S and the battery really struggles now especially when using GPS.


This happened to me. I bought a Mophie Juicepack for it and it solved the problem...by doubling the weight/volume of the phone.

Got an iPhone 6 on friday.

Yesterday I went on a 2.5 hour, 30 mile bike ride, and the iphone6 battery meter was essentially still at 100%. My 4s would have been well under 50%. Of course, the M8 is why, but after 3 years of recharging the 4s every night the battery was really starting to feel it.


Mine is the same age. Note I haven't measured but but hour plus browsing sessions have a fairly negligible impact on the battery so I'm confident it would at least reach about 5 hours (in percentage terms closer to 10 hours than 1-2).

When browsing on weak LTE the situation is different and 1-2 hours sounds very plausible. Maps/navigation can take big chunks out of the battery in a period of an hour or two and I need +60% before I set out on a 3 hour bike ride with Cyclemeter on.


Early iphone 5 units had a bad run of batteries. There is a free replacement program. Apple has a web site to check your serial number to see if your phone is afflicted. It made a world of difference on mine.


If your phone has had 2+ years of continuous use, you have definitely lost battery capacity. It does happen gradually.


Seems like Apple really knows what consumers want. I spend about 90% of my iPhone time in Safari/Chrome. Advanced 3D graphics are nice and all but I rarely take advantage of them, I'd expect that most users are the same.


I'd be interesting if those benchmarks were re-witten make use of the new Metal graphics API

https://developer.apple.com/metal/


Highly unlikely, at this point. They're commercial benchmarks, and the Metal APIs have been under beta NDA until a week or so ago.


If they weren't, would performance (score, battery usage) go even higher?


The top apps include many graphic heavy games, your usage is most likely not typical (which should be fairly obvious considering what site you're on).

Currently top grossing for the US: Clash of Clans, Candy Crush Saga, Pandora, Game of War - Fire Age, Farm Heroes Saga.


I'm not certain there is a direct correlation between the apps that make the most money in the App Store, and the apps that people spend the most time in.

It would mean that all the built in apps, and all the free apps in the App Store relatively unused, and that advertising models are a fools errand as no-one spends enough time in the apps. I suspect companies persuing these kinds of apps, including apple, are seeing enough usage to make a business case for their actions.


Of the free apps in the US store right now are keyboards (new for iOS 8 and don't really count as "apps") and Facebook Messenger, iTunes U, Amazing Thief (game), The Maze Runner (game), Instagram, Facebook, Beach Buggy Racing (Game), etc.

Games are extraordinarily popular on iOS. I see people gaming constantly on planes, trains, busses, etc.


You're assuming, however, that the top downloaded apps represent top phone usage. I have no data on it, but I doubt that's the case. Like Igglyboo, I suspect most users spend most of their phone battery on the "boring" stuff, like listening to music, looking at web pages and email.

edit: I'd also like to point out that there are two interesting metrics here: time and battery usage. It's possible for me to spend the majority of my time on boring stuff, but the majority of my battery on non-boring stuff. I'm guessing, though (again, without data and purely based on my own biases) that most people spend most of their time and battery on the boring stuff.


Are those really graphically heavy?

When I think graphically heavy I think of games like Infinity Blade. Candy Crush has extremely simple 2D graphics, same with Farm Heroes Saga and Clash of Clans.


Top downloaded apps are not the same as top used apps. I'd be willing to bet that many users are completely oblivious to (or simply dont care about) the app store and are perfectly satisfied with the default iPhone apps, I know both of my parents are.


I doubt it, the iphone is the handheld console of today, lots of people use that more than the browser. It's a shame that there's not really any good games for the iphone though (I'm looking at you, every IAP game ever).


How is it the handheld console of today when there are so few good games on the platform? I personally find the touchscreen a terrible input device for most game genres and play styles.

I can't imagine Smash Bros. on a phone. Not unless the phone gets analog sticks, face buttons, and shoulder triggers. Maybe there will be a shell that gets wide support, but given the fact that cases come in so many form factors and that Android phones have no standard shape, I doubt that will happen.

I don't mean to be negative, but phones utterly suck at the kind of games I enjoy most. 2048 is genius, but it's a phone thing.

Does anyone foresee a non-casual, non-contol limited future for mobile gaming? What changes could give it the ability to compete with the 3DS and Vita?


>iphone is the handheld console of today

>there's not really any good games for the iphone

You're contradicting yourself unless you're saying that you personally don't like any of the games.


Infinity Blade, Monument Valley and Year Walk are 3 I've played that have been frankly excellent. Year Walk was without a doubt one of my favourite games of 2013.


I don't do a lot of gaming on my phone or on a tablet, so seems to me that the best bet right now (for me) is an iPhone 6 Plus.


OT: AnandTech founder Anand Shimpi retires from journalism to work at Apple: http://arstechnica.com/gadgets/2014/09/anandtech-founder-ana...


I wonder if it's time to start scaling mobile benchmarks by the size of the device.


Doesn't the "offscreen" test do that? i.e. render something on an offscreen, fixed and uniform resolution on all devices?

Although battery performance should probably be normalized to battery size / amperage, yeah. Bigger phones usually have the same resolution as their smaller siblings but larger battery capacity.


Split up tablet and phone benchmarks but scaling by size is silly. Is it device size? Screen size? Screen resolution? Weight?


Some benchmarks are already scaled by screen size. Those have on screen and off screen versions.


It would make sense, bigger device allows for larger die and more ram, battery, etc.

Seems like bigger devices will usually squander their extra resources however and have similar performance to the smaller devices.


Do we know how Apple's Cyclone shows roughly 4x better IPC against Krait at Sunspider benchmark? Are the extra registers and bigger L1 enough to explain that?


My best guess is that the Sunspider benchmark like single-threaded performance, and Cyclone destroys Krait/Cortex here.

As for why, Cyclone's core are bigger, wider compared to Krait/Cortex, and are Out Of Order. (This is a nice article on that: http://www.extremetech.com/computing/179473-apples-a7-cyclon...)


Shouldn't this included iPad too comparisons too, since it includes the Nvida shield? Else, the Nvida shield should probably not be up there.


Nvidia made quite interesting SoC, that's why they placed Shield in the comparision. iPad has only slightly boosted iPhone SoC.


Can someone explain where there is such a huge gap between the 6 and 6+ when it comes to the Onscreen GFXBench test?


Presumably because the 6+ has a much larger screen.


Made even worse by the fact that it renders at 2208x1242 and then downsamples to 1920 × 1080.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: