Tom's Guide: A11 Bionic is the real deal (iPhone 8 review)

[Drak3]

2 hours ago, suicidalfranco said:

android can be cut down to just the necessary for the specific device. pyo

google doesn't serve a giant blob with the drivers for all devices currently available. pyo

that's what fully customisable mean. pyo

poem being lazy and not doing it doesn't mean android is soar bloated and therefore unable to be fully optimised for a specific hardware. pyo

Pyo is my thing, pyo.

 

Might I suggest Nya, Nyo, Nyu, or Gema?

[Potato*Salad]

All this useless shit in phones these days.

[captain_to_fire]

3 hours ago, TheReal1980 said:

-snip-

While this is good, I'm still not surprised since Apple controls both hardware and software. It makes me think that Apple surely does have a plan B of making their own processors for their Macs in the event that Intel becomes way too lazy and flops. 

 

3 hours ago, TheReal1980 said:

What about real-world performance? We edited and exported the same 2-minute 4K video clip on the iPhone 8, Galaxy Note 8 and Galaxy S8 using the Adobe Premiere Clip app, and it wasn’t even close. The iPhone 8 finished the task in 42 seconds, compared with more than 3 minutes for the Galaxy Note 8 and more than 4 minutes for the Galaxy S8+.

I mean, even FCP X on a 2015 MacBook Pro can outperform a 6700K with Adobe Premiere. In the case of the iPhone, Apple has more control on it than their Macs.

 

[Doobeedoo]

So OS difference on software side optimizations, hardware accelerated work hmm. NVMe and HEVC along. 

Though that chip is great. The new GPU they made and their CPU core cluster. 

[ScratchCat]

16 hours ago, Zodiark1593 said:

This is very likely. Apple does the exact same thing in Final Cut Pro by utilizing Quicksync. Exports on the Macbook (non pro) are comparable or better than a Windows workstation with Premiere.

 

Considering its virtually a given for Apple SoCs to include HEVC encoding hardware, I would be surprised if the software didn't leverage it.

 

As Apple uses NVME and some extremely fast encryption/decryption hardware, they've had the edge for awhile in opening large apps and games.

You have hit the point exactly. My atom based tablet can out perform my PC by up to 3X when encoding by using quicksync vs software H264 while consuming 3W Instead of 45W. There would be no way Apple has is able to develop a chip which out performs the competition by 3-4X using half as many cores (not counting low power cores).

 

It is however undeniably powerful , to a point where a MacBook would be very likely to appear with it or the A11X.

[Sauron]

16 hours ago, crysilis said:

the one thing Apple will always do better than android.

androids greatest feature is its biggest flaw.

its too customizable to be optimized to apple levels

apple products are optimized and made to be as user-friendly as possible

First of all performance (which is what is being discussed here) and user friendliness are two completely separate things. Secondly, os optimization can only account for a small part of any performance edge. In this case I would tend to blame Adobe for optimizing their software specifically for apple's chip and no bothering with the more varied competitors - it seems clear to me from the other tests that the difference in performance on that one test is unjustifiable from a raw power standpoint. Plus, there is something they didn't take into account.

 

I see a lot of flawed methodology; firstly by using geekbench at all and doing the usual, stupid comparison with intel hardware. Secondly, the test pool in the other areas is way too small. Third, and most glaringly obvious mistake in my opinion, they didn't take storage speed into account, at all. Loading time speedups are much more dependant on how fast the storage is and we know that the iphone probably has the fastest of the bunch, by a not small margin. This would also help justify the massive difference in editing because guess what, video editing programs have to write their output somewhere while they render and the storage drive is often slower than the cpu, particularly when it comes to modest edits like what premiere clip allows you to do. This difference becomes more obvious as you go up in resolution (and bitrate as well as file size).

 

With all that said, I don't think any of this matters. Phones have been fast enough for 99% of use cases for years now. If you really need to edit mediocre 4k footage, which you took with your phone, on your phone, then I guess the iphone has something to offer; otherwise that performance will go completely wasted and your money will go straight in the trash, because aside from that there are plenty of much cheaper phones which offer pretty much the same features if not better.

 

My suggestion is that if you're serious about video making you should save money on your phone and use it to buy a real camera, as well as do the editing on a pc.

[Sauron]

14 hours ago, hey_yo_ said:

I mean, even FCP X on a 2015 MacBook Pro can outperform a 6700K with Adobe Premiere. In the case of the iPhone, Apple has more control on it than their Macs.

Don't make the mistake of comparing two different programs to show "optimization". Any comparison of this kind MUST be made with the same exact program to be even remotely reliable. I'm sure I could take movie maker and render faster than with premiere, but I'm sure you see the problem with that reasoning.

[Tataffe]

6 minutes ago, Sauron said:

With all that said, I don't think any of this matters. Phones have been fast enough for 99% of use cases for years now. If you really need to edit mediocre 4k footage, which you took with your phone, on your phone, then I guess the iphone has something to offer; otherwise that performance will go completely wasted and your money will go straight in the trash, because aside from that there are plenty of much cheaper phones which offer pretty much the same features if not better.

That is with a big "for now". Keep in mind that iPhones have a long OS support lifespan (up to 5 years? idk what the oldest device with iOS 11 support is currently), so even future increases in computation power demand (both from apps and the OS) won't have a huge impact. This is of course an area where most Android phone makers do not even try to compete so I guess if you're one of those people who get a new phone every two years you can go for the cheapos ...

[dexT]

14 hours ago, Drak3 said:

Pyo is my thing, pyo.

 

Might I suggest Nya, Nyo, Nyu, or Gema?

That's what happens, pyo, when something is overused or put on display. Pyo. Somethings, pyo, have more impact and personal meaning if not flaunted. Pyo...

 

Something something Cultural Appropriation

 

[Sauron]

12 minutes ago, Tataffe said:

That is with a big "for now". Keep in mind that iPhones have a long OS support lifespan (up to 5 years? idk what the oldest device with iOS 11 support is currently), so even future increases in computation power demand (both from apps and the OS) won't have a huge impact. This is of course an area where most Android phone makers do not even try to compete so I guess if you're one of those people who get a new phone every two years you can go for the cheapos ...

Further increases... like what? Give me one reason for the OS to require more power in the next 5 years. Or a reason to buy this hoping it will "last" longer when for that price I could buy 3 phones over those 5 years and end up with a faster handset in the end.

 

-edit-

and by the way, in the last few years android performance degradation has been much less of a problem. For example, my oneplus x has barely lost any performance in the 2 years I've had it despite a full os upgrade and a relatively older chip. The only real difference I could detect was in boot time, but I almost never turn it off like most people.

 

An iPhone from 5 years ago would be the iPhone 5, which nowadays is outdated in pretty much everything. Sure, you can still use it, but 5 years later you're not really getting much value from the 800$ you may have spent on it in 2012.

[Tataffe]

Just now, Sauron said:

Further increases... like what? Give me one reason for the OS to require more power in the next 5 years. Or a reason to buy this hoping it will "last" longer when for that price I could buy 3 phones over those 5 years and end up with a faster handset in the end.

It is actually very unusual for the CPU power demand to not change or even decrease from one version to the next. Most of these usually stem from added background tasks or increased work of those. I'm not an expert on the matter, but you can find lots of reports saying that devices' performance decreased when users updated the OS. Glad you didn't question me mentioning app updates here as well.

From my experience, it is virtually impossible to come out cheaper when buying multiple cheapo phones instead of one flagship. 3 phones over 5 years can (again, my experience) simply not be done with cheapos since those usually last more like months rather than years. Especially if you want to have a similar phone performance experience. It simply doesn't add up. Also it's not as sustainable, but that's a whole different story.

[Sniperfox47]

Would be nice to see an Apple A11 vs Nvidia X2 comparison. Wonder where they fall relative each other. I know the DenverV2 cores are absolute beasts, and the Pascal based iGPU in the X2 is a monster as well..

 

But yeah, most of these numbers are mostly meaningless.

 

The load time for the game has more to do with storage overhead than anything... and was it the first time running the game immediately after install on Android, or did ART actually have a chance to compile it, since that could make a pretty massive difference...

 

The encoding time is pretty clearly something else at play too. Apple's Monsoon and Mistral cores may be faster than Kryo, but not 6x faster... (2 Monsoon and 4 Mistral cores vs 8 Kryo cores finishing in 1/4 the time)

[Coaxialgamer]

18 hours ago, coolkingler1 said:

Surely the optimization between hardware and software plays a big part as well? 

Possibly. That could affect the numbers a bit. But video editing involves a lot of basic number crunching, and is generally very scalable with thread count. 

Maybe the galaxy phone loses a couple percent, but that's nowhere near enough to make the difference. 

[Sauron]

3 minutes ago, Tataffe said:

It is actually very unusual for the CPU power demand to not change or even decrease from one version to the next. Most of these usually stem from added background tasks or increased work of those. I'm not an expert on the matter, but you can find lots of reports saying that devices' performance decreased when users updated the OS. Glad you didn't question me mentioning app updates here as well.

I know they do decrease, my question however is why. There is no real reason to add so many background tasks. Big additions like, say, siri are pretty superfluous in my opinion and even assuming they are necessary, those are very rare. The answer is that the developers either intentionally stifle the device's performance to force it into obsolescence, or they push the update without caring to check how it performs in older devices; this is not something you solve with more powerful hardware, because 2 years from now whatever you put in it will be obsolete anyway. It's also worth noting that you're not forced to update to the latest OS; as long as your phone is getting security updates it will be just fine with the OS it came with, or one version later.

9 minutes ago, Tataffe said:

From my experience, it is virtually impossible to come out cheaper when buying multiple cheapo phones instead of one flagship. 3 phones over 5 years can (again, my experience) simply not be done with cheapos since those usually last more like months rather than years. Especially if you want to have a similar phone performance experience. It simply doesn't add up. Also it's not as sustainable, but that's a whole different story.

Bear in mind that we're talking about 250$+ devices as "cheapos" here. My phone was about that price and it has worked excellently for the past 2 years, and I expect it to do the same for at least as much more time. 3 phones over 5 years is an exaggeration, but perfectly feasible and you wouldn't be frustrated by the experience. In reality it would more likely be one phone until it breaks or becomes painfully slow (hopefully not earlier than 4 years from purchase), with the best case scenario being that the phone lasts 5 years or more and you simply saved the 600$ difference - but if it DOES have problems before that, you can buy a new one and still come out with a faster device and more money in your pocket.

[Spenser1337]

Can't wait to start rendering 4K videos on my phone! Some people ask why I'm not using my computer, to that I answer "the grind doesn't stop bitch"

 

i heard 4K looks really good on the 1080p screen too

[Zodiark1593]

19 minutes ago, Tataffe said:

It is actually very unusual for the CPU power demand to not change or even decrease from one version to the next. Most of these usually stem from added background tasks or increased work of those. I'm not an expert on the matter, but you can find lots of reports saying that devices' performance decreased when users updated the OS. Glad you didn't question me mentioning app updates here as well.

From my experience, it is virtually impossible to come out cheaper when buying multiple cheapo phones instead of one flagship. 3 phones over 5 years can (again, my experience) simply not be done with cheapos since those usually last more like months rather than years. Especially if you want to have a similar phone performance experience. It simply doesn't add up. Also it's not as sustainable, but that's a whole different story.

Cheap Android phones have been on the same Cortex A53 architecture for a few years now, which just for the OS itself, runs just fine.

[79wjd]

1 hour ago, ScratchCat said:

There would be no way Apple has is able to develop a chip which out performs the competition by 3-4X using half as many cores (not counting low power cores).

Edit: MisreadWhat high end mobile ARM CPUs are using 6-8 high power cores? The SD835 and Exynos 8895 both only have four high power cores. And look at Geekbench (which I know isn't a great benchmark), and you can argue that optimization between the two platforms means the scores aren't directly comparable, but I find it hard to believe that GB is so much more optimized on iOS that the optimization accounts for the entirety of the very large difference between scores. 

• The iPhone 8 scores: around 4200, 10,000.
  • The iPhone 7 scores around 3500, 6000.
• The SD835 Note 8 scores around 1800, 6000.
  • The 820 S7 scores around 1600, 3800.
• The 8895 S8 scores around 1950, 6500.
  • The 8890 S7 scores around 1800, 5300.

1 hour ago, Tataffe said:

iPhones have a long OS support lifespan (up to 5 years? idk what the oldest device with iOS 11 support is currently), 

iOS 11 supports up to the iPhone 5s (not the 5/5c), but historically iOS has been supported on five year old devices, Apple just decided to move to just 64 bit only with iOS 11 and the 5/5c are only 32 bit. 

 

[mawv]

17 hours ago, coolkingler1 said:

Dude don't pyo everything.

What's this whole pyo thing

On topic, Apple does it again.

[AlwaysFSX]

14 minutes ago, mawv said:

What's this whole pyo thing

On topic, Apple does it again.

For when you're trying to be obnoxious because not being obnoxious on the internet is too hard.

 

"Pyo" is a word that an adult female who frequents forums, constantly, uses in place of "Stop" when typing a message. It's considered a Very Annoying "Catch Phrase" anywhere the user goes.

Piyoko: Mine was a hundred bucks, it could have been cheaper, but I didn't feel like waiting for shipping, and wanted a nicer one, pyo. 

Piyoko: Oh, and thanks, pyo.

[mawv]

Just now, AlwaysFSX said:

For when you're trying to be obnoxious because not being obnoxious on the internet is too hard.

 

"Pyo" is a word that an adult female who frequents forums, constantly, uses in place of "Stop" when typing a message. It's considered a Very Annoying "Catch Phrase" anywhere the user goes.

Piyoko: Mine was a hundred bucks, it could have been cheaper, but I didn't feel like waiting for shipping, and wanted a nicer one, pyo. 

Piyoko: Oh, and thanks, pyo.

Christ...

[AlwaysFSX]

10 minutes ago, mawv said:

Christ...

For the people who try too hard to be meta, that's the thing apparently.

[Sniperfox47]

7 minutes ago, djdwosk97 said:

What high end mobile ARM CPUs are using 6-8 high power cores? The SD835 and Exynos 8895 both only have four high power cores. 

SD835 has 8 full power Kryo cores.

 

And the A11 only has 2 Big (high power) cores and 4 little (efficiency) cores. So even the Exynos 8895 has twice the Big cores.

 

10 minutes ago, djdwosk97 said:

And look at Geekbench (which I know isn't a great benchmark), and you can argue that optimization between the two platforms means the scores aren't directly comparable, but I find it hard to believe that GB is so much more optimized on iOS that the optimization accounts for the entirety of the very large difference between scores. 

• The iPhone 8 scores: around 4200, 10,000.
  • The iPhone 7 scores around 3500, 6000.
• The SD835 Note 8 scores around 1800, 6000.
  • The 820 S7 scores around 1600, 3800.
• The 8895 S8 scores around 1950, 6500.
  • The 8890 S7 scores around 1800, 5300.

Geekbench isn't just a cpu benchmark though, and Apple also has kick-ass storage. It's also a pretty lightly threaded test, so the fewer but more powerful cores in the iPhone do better.

[79wjd]

9 minutes ago, Sniperfox47 said:

SD835 has 8 full power Kryo cores.

 

And the A11 only has 2 Big (high power) cores and 4 little (efficiency) cores. So even the Exynos 8895 has twice the Big cores.

 

Geekbench isn't just a cpu benchmark though, and Apple also has kick-ass storage. It's also a pretty lightly threaded test, so the fewer but more powerful cores in the iPhone do better.

Aren't four of the cores in Kryo 280 downclocked by a decent amount? 

EDIT: The SD835 is still running a variation on big.LITTLE with four cores clocked at 2.45ghz and four cores clocked at 1.8ghz (a 40%~ difference).

 

2x as many high power cores in the 8895 is still far from 3-4x as many high power cores. And 2x is what my post implied besides. 

 

I also listed the single threaded results as well for that very reason. 

[Sniperfox47]

1 minute ago, djdwosk97 said:

2x as many high power cores in the 8895 is still far from 3-4x as many high power cores. And 2x is what my post implied besides.

The guy you quoted said that they have 2x the number of cores of Apple, not 3-4x... They said Apple doesn't have 3-4x from just those 2x the cores.

 

2 minutes ago, djdwosk97 said:

I was also looking at just the single threaded numbers as well. 

...why are you looking at single threaded performance in an argument of the relative performance of different numbers of cores?

[79wjd]

10 minutes ago, Sniperfox47 said:

...why are you looking at single threaded performance in an argument of the relative performance of different numbers of cores?

Your argument against geekbench is that it's lightly threaded even on the multi threaded test. If you look at the single threaded performance of each CPU and then scale it up to the number of available cores, then you eliminate that issue. (Of course, that creates a whole bunch of other issues, but regardless, it is another datapoint to estimate multithreaded performance)