• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Anandtech releases preliminary iPhone 6 Perf/Battery tests

Status
Not open for further replies.
Kysen said:
It drives home the point that Apple being in control of both hardware and software is what allows it to stay in the game. I get the feeling that Android is the weak link for all those other phones. No point having all that power under the hood if the OS itself is wasteful.

The geekbench tests also show a similar pattern to the intel vs amd divide. More cores in the competition but slower singled threaded performance and everything suffers because of this design.
Click to expand...

The biggest thing is that most of these 2+ GHz Krait designs are pushing past 4W and the typical phone chassis has trouble dumping that much heat. You can't really let the chip go all out on sustained multitasked workloads because it gets thermally limited within 5-10 minutes. Like take this from the Nexus 4. It's a screenshot of an N4 running NFS:MW after 5 minutes.

Screenshot_2013-11-19-17-23-11.jpg
Click to expand...

It's already scaled back to 2 cores running at 1.2GHz after starting at 1.5GHz!

Apple went really wide on the A7/A8 to improve single threaded performance significantly compared to the stock A53/A57 designs. Like ridiculously wide. But Apple also control the entire toolchain, write their own APIs, write their own god damned compiler. So the end result is that they can wring out every last bit of it that they can and it shows. Physics performance sucks because it only has two cores but web browsing benchmarks are stellar.
 
My iPhone 5 lasts about 3-4 hours on a good day with normal use. I qualify for the battery replacement program. Just have to drive over 100 miles to an Apple Store which sucks. My 5 had great battery life in the beginning too. Hopefully they learned their lesson this go around.
 
why are Toms Hardware benchmarks for ipHone 6 and Plus so different than Anandtech?

Why does Anandtech result show iPhone 6 plus have a much lower Manhattan benchmark score but Toms hardware has it as equal?

http://www.tomsguide.com/us/iphone-6-benchmark-results,news-19584.html

j9GhX2h.jpg


0SYUX3e.jpg


When also ran the Manhattan 1080p offscreen test, which uses the same 1080p resolution on every device. The iPhone 6 Plus still came out on top with a score of 18.7 fps. Next was the iPhone 6 (17.8 fps), the iPhone 5s (12.9 fps) and the Samsung S5 (11.9 fps).
Click to expand...


yK2Teg4.jpg


gzhURP7.jpg


Rx3m9MU.jpg


xg4C7G5.jpg


RPYpuyq.jpg


s4wbJrP.jpg
 
Sir Abacus said:
The biggest thing is that most of these 2+ GHz Krait designs are pushing past 4W and the typical phone chassis has trouble dumping that much heat. You can't really let the chip go all out on sustained multitasked workloads because it gets thermally limited within 5-10 minutes. Like take this from the Nexus 4. It's a screenshot of an N4 running NFS:MW after 5 minutes.



It's already scaled back to 2 cores running at 1.2GHz after starting at 1.5GHz!

Apple went really wide on the A7/A8 to improve single threaded performance significantly compared to the stock A53/A57 designs. Like ridiculously wide. But Apple also control the entire toolchain, write their own APIs, write their own god damned compiler. So the end result is that they can wring out every last bit of it that they can and it shows. Physics performance sucks because it only has two cores but web browsing benchmarks are stellar.
Click to expand...
great summary. Yeah, apple is holding off on lengthening their cores until 14nm which likely(?) should hit next year. 14nm should allow them to lengthen the pipeline and hit higher clock speeds while not running into a major wattage or thermal issue.

The designs of Apple's CPUs should not be underestimated.. and bringing Anand onboard was likely only done to strengthen that position. Next year's CPUs (provided that's when they'll go to 14nm) should be pretty insane.
 
Maninthemirror said:
why are Toms Hardware benchmarks for ipHone 6 and Plus so different than Anandtech?

Why does Anandtech result show iPhone 6 plus have a much lower Manhattan benchmark score but Toms hardware has it as equal?
Click to expand...

Onscreen renders at the phone's native resolution, i6+ has a higher resolution so more pixels to push so lower FPS. Offscreen renders at 1080p no matter what so it taxes the graphics hardware in the exact same way based purely on specs.
 
Kysen said:
Holy shit that's bad. So in that case falling back to less cores at even lower speeds compounds the issue.
Click to expand...

It's not that terrible because most games don't really tax the CPU as much as you would think. But you can gain 20-50% in benchmarks just by telling the SoC to ignore thermal limitations while the benchmark is going.

*cough*Like Samsung have done in the past*cough*
 
I haven't really paid that much attention in the past but I'm curious about how they setup the phones for test.

Do they test it in factory state or do they test with a user signed in? The battery performance drops quite a bit more on Android due to background tasks once you have a Google user signed in.

edit: my idling N7 without a Google user signed in can last for about month as oppose to barely 1 week. So that would affect the 'browsing battery test' quite a bit depends on how it's setup.
 
Sir Abacus said:
Apple went really wide on the A7/A8 to improve single threaded performance significantly compared to the stock A53/A57 designs. Like ridiculously wide. But Apple also control the entire toolchain, write their own APIs, write their own god damned compiler. So the end result is that they can wring out every last bit of it that they can and it shows. Physics performance sucks because it only has two cores but web browsing benchmarks are stellar.
Click to expand...

Most people consider Apple a hardware company, but in reality they have a really, really, really good (and versatile) software foundation to build on. (Something they were sorely lacking in the '90s.)

Endless bitching about the UI aside, it's the stuff people don't see that they forget about.
 
jstripes said:
Most people consider Apple a hardware company, but in reality they have a really, really, really good (and versatile) software foundation to build on. (Something they were sorely lacking in the '90s.)

Endless bitching about the UI aside, it's the stuff people don't see that they forget about.
Click to expand...

this is also the case with people upgrading from 7 to 8 and people say, whats the change? I dont see any changes, not realising that in reality iOS 8 is a bigger change than iOS 6 to iOS 7 because it completely changes the layer underneath the UI
 
Sir Abacus said:
It's not that terrible because most games don't really tax the CPU as much as you would think. But you can gain 20-50% in benchmarks just by telling the SoC to ignore thermal limitations while the benchmark is going.

*cough*Like Samsung have done in the past*cough*
Click to expand...

Even if it's not just for the benchmark, running for a 1 minute long benchmark will often not cause the phone to hit normal thermal limits. Environment has a massive impact on what these phones hit in terms of benchmarks. Try running a Galaxy S5 at max brightness (which, because it's an OLED screen makes it super hot on its own) and play a taxing game for 10 minutes then do your benchmark... you'd see the results tank due to the phone either being on fire or throttled to hell.
 
Maninthemirror said:
this is also the case with people upgrading from 7 to 8 and people say, whats the change? I dont see any changes, not realising that in reality iOS 8 is a bigger change than iOS 6 to iOS 7 because it completely changes the layer underneath the UI
Click to expand...

Or the iPhone 4 to 4S, iPhone 5 to 5S.

"It looks the same. WTF?"
 
Sir Abacus said:
The biggest thing is that most of these 2+ GHz Krait designs are pushing past 4W and the typical phone chassis has trouble dumping that much heat. You can't really let the chip go all out on sustained multitasked workloads because it gets thermally limited within 5-10 minutes. Like take this from the Nexus 4. It's a screenshot of an N4 running NFS:MW after 5 minutes.



It's already scaled back to 2 cores running at 1.2GHz after starting at 1.5GHz!

Apple went really wide on the A7/A8 to improve single threaded performance significantly compared to the stock A53/A57 designs. Like ridiculously wide. But Apple also control the entire toolchain, write their own APIs, write their own god damned compiler. So the end result is that they can wring out every last bit of it that they can and it shows. Physics performance sucks because it only has two cores but web browsing benchmarks are stellar.
Click to expand...

On that note, I thought it was really interesting - and telling - that Apple called specific attention to this during the iPhone 6 announcement. They made a point of showing how the A8 can run at full blast with no throttling for a much longer length of time than the A7 or most other phones:

A8-07.jpg


I'm sure Anandtech will do their own tests, but Apple basically implied the A8 can go for up to twenty minutes at full load without throttling, which is absolutely crazy in the smartphone world.
 
rocksteady1983 said:
My iPhone 5 lasts about 3-4 hours on a good day with normal use. I qualify for the battery replacement program. Just have to drive over 100 miles to an Apple Store which sucks. My 5 had great battery life in the beginning too. Hopefully they learned their lesson this go around.
Click to expand...

Dude, mail the phone in to them. I think they might even send you a box for it.

soundahfekz said:
Weird not to include the Xperia Z3c or Z3, which shits on everything battery life wise.

Great optimization though by Apple.
Click to expand...

Aren't Sony's phones basically irrelevant from a sales standpoint? Guessing that's why they weren't included.
 
Guess Who said:
On that note, I thought it was really interesting - and telling - that Apple called specific attention to this during the iPhone 6 announcement. They made a point of showing how the A8 can run at full blast with no throttling for a much longer length of time than the A7 or most other phones:

I'm sure Anandtech will do their own tests, but Apple basically implied the A8 will go for twenty minutes at full load without throttling, which is absolutely crazy in the smartphone world.
Click to expand...

phone wars aside, apple SHOULD be proud of Cyclone. It is still the only true 64-bit ARM implementation first to the punch and is just an amazing mobile CPU. Makes sense that they want to brag about it when they can. It also sets up (probably) huge expectations for Cyclone's successor.

edit - By "true" I mean that I believe the only other 64-bit mobile implementations out on the market are one with two quad core ARM CPUs that are both still 32-bit but will accept and run ARMv8-A instructions while handling them on the 32+32 architecture. If this is wrong someone feel free to correct me.
 
Guess Who said:
On that note, I thought it was really interesting - and telling - that Apple called specific attention to this during the iPhone 6 announcement. They made a point of showing how the A8 can run at full blast with no throttling for a much longer length of time than the A7 or most other phones:

I'm sure Anandtech will do their own tests, but Apple basically implied the A8 can go for up to twenty minutes at full load without throttling, which is absolutely crazy in the smartphone world.
Click to expand...

You're running a 20nm dual core processor at 1.4GHz. A7 was already peaking at about 2.6W to go flat strap. I suspect the A8 dragged that down to 2W combined with a case that's literally a heatsink.
 
I'm actually most impressed by the HTC One E8. This is supposed to be a budget version of the M8 :O
 
I skipped around a lot on the keynote video and didn't realize they talked about sustained performance. it's a decent point especially considering how big gaming is on these things.

i'd like to see Anand benchmarks showing a difference between 5S and 6 throttling.
 
Kysen said:
It drives home the point that Apple being in control of both hardware and software is what allows it to stay in the game. I get the feeling that Android is the weak link for all those other phones. No point having all that power under the hood if the OS itself is wasteful.

The geekbench tests also show a similar pattern to the intel vs amd divide. More cores in the competition but slower singled threaded performance and everything suffers because of this design.
Click to expand...

It has nothing to do with the OS, and more to do with the methodology behind the processors.

Apple's CPUs have huge single core performance, but the caveat is that have a low upper performance limit. The chip has to basically be re-engineered every year to get more processor speed out of it. It is similar to the Wii U CPU in that regard, where it has huge performance for it's perceived specs, but is basically useless for scale. nVidia and Qualcomm build for scale so that they do not have to re-engineer their process too often. It is much easier and cheaper to add cores than it is to bump the frequency in a short pipeline. It has absolutely nothing in common with the Intel vs AMD debate. Intel's designs are not held back by limits, they are held back because they are obliterating the competition and have nothing to gain by extending that gap.
 
Cheech said:
Dude, mail the phone in to them. I think they might even send you a box for it
Click to expand...
Theres no way I can be without a phone for that long. Unless they just send me a complete phone replacement and I mail the old phone back. How do you even go about that. The site only mentions taking your phone in person to a store.
 
My Lumia 1520, last me almost 2 days and blows any Android phones in terms of battery and its not even on the list. Joke comparision
 
bitocoder said:
It has nothing to do with the OS, and more to do with the methodology behind the processors.

Apple's CPUs have huge single core performance, but the caveat is that have a low upper performance limit. The chip has to basically be re-engineered every year to get more processor speed out of it. It is similar to the Wii U CPU in that regard, where it has huge performance for it's perceived specs, but is basically useless for scale. nVidia and Qualcomm build for scale so that they do not have to re-engineer their process too often. It is much easier and cheaper to add cores than it is to bump the frequency in a short pipeline. It has absolutely nothing in common with the Intel vs AMD debate. Intel's designs are not held back by limits, they are held back because they are obliterating the competition and have nothing to gain by extending that gap.
Click to expand...

true... but the move to 14nm "should" allow them to lengthen out the pipelines a bit letting them ramp up speed as well.. but yeah like you said, at basically a re-engineering (both the CPU and LLVM)

if 14nm weren't on the horizon apple would likely have a rough go of it next year.
 
borghe said:
great summary. Yeah, apple is holding off on lengthening their cores until 14nm which likely(?) should hit next year. 14nm should allow them to lengthen the pipeline and hit higher clock speeds while not running into a major wattage or thermal issue.

The designs of Apple's CPUs should not be underestimated.. and bringing Anand onboard was likely only done to strengthen that position. Next year's CPUs (provided that's when they'll go to 14nm) should be pretty insane.
Click to expand...
Nope. 2016 at the earliest unless apple goes Intel for the 6S, which is highly unlikely.


While the iPhone 6 battery life results are impressive, remember that its on a newer process node than everything else on the chart with much more mature display tech and a much more capable gpu. Yes, optimization plays a large role, and that's one thing apple deserves credit for. Their hardware/software unison is unparalleled. However, once the other OEMS jump to 20nm we should start seeing improvements across the board. What Sony has done with the Z3 and Z3c is just as impressive as what apple has done with the 6. When they jump to 20nm, we should be eeking out 10 hours of screen time in real world usage.

One more thing. These number do not indicate real world usage, and saying "I got 40 hours out of my iPhone 6, it's amazing" without screen time numbers isn't really saying much.
 
malyce said:
Nope. 2016 at the earliest unless apple goes Intel for the 6S, which is highly unlikely.
Click to expand...

It would be a hell of a task but not unprecedented. Panasonic, Tabula, Netronome, Microsemi, Altera and Achronix have all licensed 14nm and have access to Intel's foundry capacity.
 
sirap said:
All iphones batteries start out strong, then quickly degrade in the following months. My 5s had amazing battery life too, now I can barely keep it on. Maybe I'm just charging them wrong, but the same thing happened to my 5, 4s, 4 etc.

Here's hoping the 6 Plus's battery is easily replaceable.
Click to expand...

It'll do the same. My iPhone 5 battery was amazing, then after iOS updates it just went to junk. Losing juice rapidly while on anything including homescreen.
 
Fularu said:
This is much more realistic. Especially regarding my own experience with my iPhone 5.
Click to expand...

I actually think there are weird results both on Phone Arena's test and Anandtech's test actually.

Anticipating GSMArena's since the Xperia Z3 Compact's test results was absolutely nuts. Between all three websites, we will get a general idea of what we'll see. My HTC One M7 battery life general fell in line with the results of the GSMarena test even though I was on LTE most of the day instead of wi-fi.

I never trust GSMArena's results for Nexus devices because they never retest after the first new Android version release, which always offers much better battery life.
 
How are these battery tests conducted? That can't be with cell network enabled right? Seems far too high. My HTC One M7 comes no where close to that.

Reminds me how poor my HTC One battery has become the last few months. Wakelocks out of control and really difficult to manage. Way too hard to find the apps causing them and then fix them (even with wakelock detector apps). :/
 
So what I'm getting is that my 5S is actually still pretty competitive

Except besides the iphone 6+ which seems to be a beast battery wise
 
Fularu said:
This is much more realistic. Especially regarding my own experience with my iPhone 5.
Click to expand...

Not based on my experience with the nexus 5.. No way does the iphone 6 only gets 30 minutes more battery life. I have 52% left, where my Nexus 5 was dead 3 hours ago...

Edit: I don't trust this review at all based on my experience with the iPhone 6 (so far) and my experience with the Nexus 5 for almost a year. They aren't even close.
 
Cheech said:
Dude, mail the phone in to them. I think they might even send you a box for it.



Aren't Sony's phones basically irrelevant from a sales standpoint? Guessing that's why they weren't included.
Click to expand...

The fuck does sales have to do with benchmarks, which is what this thread is about?

You iPhone dudes are killing me right now. If no rebuttal or legitimate comment to make, then SALES SALES SALES SALES


NH Apache said:
Phone Arena:



http://www.phonearena.com/news/All-...3-Compact-score-a-record-battery-life_id60922
Click to expand...


Damn, Huwei and Z3C are beasts on battery life.
 
soundahfekz said:
The fuck does sales have to do with benchmarks, which is what this thread is about?

You iPhone dudes are killing me right now. If no rebuttal or legitimate comment to make, then SALES SALES SALES SALES





Damn, Huwei and Z3C are beasts on battery life.
Click to expand...

The Z3C hasn't been finished re-testing yet. That's the Xperia Z3 result.

I was thinking that these tests have to be bullshit though because in the comments section, people were mentioning that the first time about Xperia Z3C clocked in at 14 hours and 44 mins.

But then I remembered what GSMarena tested just for the Wifi browsing portion:

battest.jpg
Click to expand...
 
reKon said:
The Z3C hasn't been finished re-testing yet. That's the Xperia Z3 result.

I was thinking that these tests have to be bullshit though because in the comments section, people were mentioning that the first time about Xperia Z3C clocked in at 14 hours and 44 mins.

But then I remembered what GSMarena tested just for the Wifi browsing portion:
Click to expand...

Yeah GSM is usually pretty on point. That amount of battery life makes it tempting for sure.

Not trying to detract from the topic at hand though, sorry guys
 
soundahfekz said:
The fuck does sales have to do with benchmarks, which is what this thread is about?

You iPhone dudes are killing me right now. If no rebuttal or legitimate comment to make, then SALES SALES SALES SALES





Damn, Huwei and Z3C are beasts on battery life.
Click to expand...

What is he trying to rebut? Are you implying Anandtech had ulterior motives for leaving the Sony phones off the benchmarks? I would think that including dozens of phones, some of which aren't marketed or sold in high numbers, wouldn't be feasible for them to do. Obviously it would be preferable and make for better benchmarks, but... I'm not seeing your reasoning for attacking anyone as being iPhone fantards.
 
Status
Not open for further replies.

Similar threads

Foldable iPhone renders
2
79
2K
Trilobit replied
Kagurabachi Anime Releases April 2027
News 
5
163
thefool replied
House of the Dragon Season 2 releases July 21 2026
News 
28
369
jason10mm replied
No, you´re not getting old: iPhone keyboard is broken
15
745
TheUnicornGuy replied
We got Linus Torvalds vibe coding before GTA 6
4
319
MidGenRefresh replied
Top Bottom