According to Toms Hardware testing the Samsung is slightly better overall lol.

[CraigGB]

im pleased you are happy with your phone, im not saying the samsung is a bad chip, just most of the evidence suggests the tsmc is better overall

personally I think it's far too early to jump that conclusion, the panic seems to have driven a couple of people insane, and really the evidence seems to show the geekbench battery test is an anomaly, so far we have:

the geekbench battery test which shows a massive difference between the two chips.

all of anandtech's benchmarks minus the geekbench battery test show the two to perform within a few percent of each other, the geekbench battery test of course shows the TSMC chip massively ahead.

tomshardware performed their own suite of benchmarks and also found the two chips within a few percent of each other.

several other sites have found the same with real world usage tests, there was a german one somewhere in one of the threads with a pretty good real world test between the two which in the end showed them to perform nearly identical and battery life was within two percent between the two, the TSMC finished at 38% and the Samsung was at 40%, a difference that could easily be down to manufacturing variance.

several video's running a variety of tasks in a real world setting show the performance to be very close, I've included the most thorough in this post that runs several CPU intensive tasks for a long amount of time, the end result is they are within 1 percent of each other in regards to battery and a minor difference in length taken to perform the tasks, again perfectly normal variations in the silicon lottery. they also run the geekbench battery test and it as usual it shows a massive difference between the two, it's possibly the best evidence I've seen that the geekbench battery test is not just an anomaly but it seems to be flawed, however I have no idea what might be causing it, hoping it's figured out soon.

apologies for the long post, honestly I'm just waiting to see what's going on. I would rather wait until we know the answers before forming a conclusion, anyway, haha it feels like I've recounted the last couple of days just about in regards to chip gate, the video I mentioned is below.

 

[Anand953]

Dont worry there will be another article next week saying the tmsc is better haha. At this point just keep your phones and enjoy it. If your getting poor battery life you may have another issue.

The best way to avoid all this crap is to just live with your phone and not check which chip oils it.. If you are facing serious issues on your phone during your warranty period just rush to apple support to get it replaced..

A fraction of speed here and and a few extra minutes of battery there is not going to make any difference in your day to day life..

It's high time for us to just chillax and enjoy our new purchases

 

[Dented]

Result from a German website:

Battery benchmark scores:

(3511-2450)/2450 = 43.3% ---- in TSMC's favor.

Images of battery test result:
http://abload.de/img/samsung_vs_tsmc_akku1os3a.jpg

The web page (In German; scroll down to see the pictures)
http://f.ifun.de/discussion/1265/6s-chipgate-test-tsmc-vs-samsung-benchmarks-screenshots-inside

How did I know it was the geekbench test before I saw the screenshot. Same story, just different language..

 

[kvnt]

The following is a CPU-intensive test, other than the Geekbench, without background programs, with minimal influence from other components such as display and wifi.

"The brightness and volume are adjusted to minimum, and to turn off automatically to reduce the influence of the screen. ... The phones are fully charged to 100% and restarted.

"Test 1: Safari Javascript rendering complex animations. ... CPU utilization is running about 60%. After 38 minutes, the Samsung 14nm has 59% charge remaining [but the] TSMC 16nm has 74%."

http://www.computerworld.com/articl...dware/apple-iphone-6s-batterygate-itbwcw.html

Based on the above test result, Samsung iphone consumes 41% of a full charge; TSMC, 26%. So, TSMC is better by

(41-26)/26 = 57.7%

Or, put it another way:

On the same CPU-intensive task, TSMC phone uses 26 units of power; while Samsung consumes 41 units.

26 * 1.577 = 41. Therefore, Samsung uses 57.7% more power.

Tentative conclusion:

Many tests, Geekbench or others, had shown Samsung A9 has far worse power efficiency, under heavy CPU workloads.

However, since components other than A9 consume most power and not everyone runs CPU-intensive programs, the total battery gap on the iphone may be only in the single digits.

 

[stulaw11]

Anyone making conclusions citing ONE test on ONE device is already wrong. Unless you take a sample of thousands of TSMC and thousands of Samsung phones, there will be no definitive conclusion. Basic statistics 101.

And even then, different benchmarks measure metrics differently- sometimes a combinations of processes which one chip may perform better than the other. Ie. TSMC may prevail in 3 processes but lose in 3 in reality making them closer than the benchmark states, depending how heavily it weighs on any one of the processes/metrics to give the score.

The conclusion is there will likely be no conclusion. Anyone making conclusions is is flat out wrong to begin with

Now run say a video on the 2 with no other apps open, not connected to cellular and wifi,same brightness and test the battery drain and MAYBE then you have some general battery tendencies in the real world, But again, only in ONE type of metric- video playback.

There's no way to really say definitely TSMC>Samsung or reverse though as everyone's real world usage and apps open/notifications/cell and WIFI signal strength is different.

 

[hammy434]

The following is a CPU-intensive test, other than the Geekbench, without background programs, with minimal influence from other components such as display and wifi.

"The brightness and volume are adjusted to minimum, and to turn off automatically to reduce the influence of the screen. ... The phones are fully charged to 100% and restarted.

"Test 1: Safari Javascript rendering complex animations. ... CPU utilization is running about 60%. After 38 minutes, the Samsung 14nm has 59% charge remaining [but the] TSMC 16nm has 74%."

http://www.computerworld.com/articl...dware/apple-iphone-6s-batterygate-itbwcw.html

Based on the above test result, Samsung iphone consumes 41% of a full charge; TSMC, 26%. So, TSMC is better by

(41-26)/26 = 57.7%

Or, put it another way:

On the same CPU-intensive task, TSMC phone uses 26 units of power; while Samsung consumes 41 units.

26 * 1.577 = 41. Therefore, Samsung uses 57.7% more power.

Tentative conclusion:

Many tests, Geekbench or others, had shown Samsung A9 has far worse power efficiency, under heavy CPU workloads.

However, since components other than A9 consume most power and not everyone runs CPU-intensive programs, the total battery gap on the iphone may be only in the single digits.

well i started to think it was just geekbench and maybe there isnt an issue, but looks like there is. im just gonna wait for more tests i think, then we'll fully know.

 

[kvnt]

well i started to think it was just geekbench and maybe there isnt an issue, but looks like there is. im just gonna wait for more tests i think, then we'll fully know.

Earlier in the year, when Samsung’s 14nm Exynos 7420 debuted, everyone cited Geekbench to label the Exynos as the top smartphone AP. Nobody said the Geekbench was flawed.

Now the Geekbench shows that Samsung A9 delivers poor power efficiency under heavy CPU load. The Geekbench must be flawed.

Only the Geekbench, which shows Samsung superiority, is a good benchmark?

 

[CraigGB]

well i started to think it was just geekbench and maybe there isnt an issue, but looks like there is. im just gonna wait for more tests i think, then we'll fully know.

if you follow the actual article which is in what I think is chinese, after he performs the test he states:

I prefer to believe this is a case of my phone, it is a small probability event, so I will go to Apple stores the replacement, hope or Samsung 14nm, so that I can test again with the same conditions, to ensure the reasonableness of the results!And then I will update here.

there is then a second test which as far as I can make of it he has a new phone, he uses no sim card, turns the brightness to the lowest and basically turns off as much as he can to try and judge battery usage mostly by CPU usage which in a sense should amplify the CPU difference, he then runs a test and by the end of it he concludes

the temperature difference of about 3 degrees, the power difference of 6%

the article then ends with

But for everyday use, it can not possibly be sustained high-load operation, ran scores we saw it, almost 2% less points difference, can understand. So I still believe that Apple's quality control, and user experience nothing much difference, which is Apple again pursued unless you run every test points, generally do not need to use the tangle is 16nm or 14nm.

the translation makes it a bit hard to understand but from what I understand he basically says unless you run every benchmark which will amplify the difference you don't need to worry about real world battery life I think?

kvnt didn't post the follow up so I did

Earlier in the year, when Samsung’s 14nm Exynos 7420 debuted, everyone cited Geekbench to label the Exynos as the top smartphone AP. Nobody said the Geekbench was flawed.

Now the Geekbench shows that Samsung A9 delivers poor power efficiency under heavy CPU load. The Geekbench must be flawed.

Only the Geekbench, which shows Samsung superiority, is a good benchmark?

you weren't on android central then, the Exynos 7420 performed great in geekbench, the Galaxy S6 performed great in geekbench with that CPU, problem was in the real world battery life wasn't anything like geekbench indicated it should be, as you might notice the Galaxy S6 is above the iPhone 6+ in the charts, in regards to battery life it is pretty poor. 3-5 hours of screen time is all you can get out of it. the geekbench battery result didn't translate into real world use at all meaning the results were indeed flawed even compared to other android phones. that annoyed quite a few people

 

[iAstonish]

if you follow the actual article which is in what I think is chinese, after he performs the test he states:

I prefer to believe this is a case of my phone, it is a small probability event, so I will go to Apple stores the replacement, hope or Samsung 14nm, so that I can test again with the same conditions, to ensure the reasonableness of the results!And then I will update here.

there is then a second test which as far as I can make of it he has a new phone, he uses no sim card, turns the brightness to the lowest and basically turns off as much as he can to try and judge battery usage mostly by CPU usage which in a sense should amplify the CPU difference, he then runs a test and by the end of it he concludes

the temperature difference of about 3 degrees, the power difference of 6%

the article then ends with

But for everyday use, it can not possibly be sustained high-load operation, ran scores we saw it, almost 2% less points difference, can understand. So I still believe that Apple's quality control, and user experience nothing much difference, which is Apple again pursued unless you run every test points, generally do not need to use the tangle is 16nm or 14nm.

the translation makes it a bit hard to understand but from what I understand he basically says unless you run every benchmark which will amplify the difference you don't need to worry about real world battery life I think?

kvnt didn't post the follow up so I did

Lol I just posted my battery life in the android thread I made and someone got mad and called me a troll because it's "impossible" to get 10 hours of screen on time. So I guess the Samsung chip is doing pretty well if it's embarking into impossible territories.

 

[kvnt]

the temperature difference of about 3 degrees, the power difference of 6%
....

I see different second test in the link. Readers can verify for themselves.

What is clear indicated is the second test:
http://www.computerworld.com/articl...dware/apple-iphone-6s-batterygate-itbwcw.html

“Test 2: NPlayer LAN RMVB video playback. ... After 106 minutes, the Samsung 14nm has 42% charge remaining [but the] TSMC 16nm has 57%.“

The original page with pictures: http://m.mydrivers.com/newsview/449771.html?ref=

So, the second test renders 58 vs. 42. TSMC advantage is 35%. This is consistent with Samsung’s problem on heavy CPU loads, as the video playback uses more GPU than CPU comparing to the the first test of 57% lead.

This a A9 only comparison, which is a gauge of the underlying process technologies. However, I had said before: “since components other than A9 consume most power and not everyone runs CPU-intensive programs, the total battery gap on the iphone may be only in the single digits."

 

[roeiz]

Lol I just posted my battery life in the android thread I made and someone got mad and called me a troll because it's "impossible" to get 10 hours of screen on time. So I guess the Samsung chip is doing pretty well if it's embarking into impossible territories.

you little devil

 

[roeiz]

This is consistent with Samsung’s problem on heavy CPU loads, as the video playback uses more GPU than CPU comparing to the the first test of 57% lead.

so what now? Video playback has that much difference?
if THAT is a definite fact, we DO have a problem.
don't get me started again!

 

[Dented]

so what now? Video playback has that much difference?
if THAT is a definite fact, we DO have a problem.
don't get me started again!

I believe the current argument is that video playback doesn't count in a benchmark as it doesn't use the cpu so much, which is why Samsung chipped phones exhibit no difference in video playback. Or video exporting. Or [insert any other activity you might actually use a smartphone for other than running Geekbench].

 

[kvnt]

the temperature difference of about 3 degrees, the power difference of 6%
....

so what now? Video playback has that much difference?
if THAT is a definite fact, we DO have a problem.
don't get me started again!

Don’t worry. That test was done with display and wifi and cellular off, in order to test A9 only. Otherwise, the battery gap will be a lot smaller.

When you watch videos on youtube, your display and wifi and cellular are definitely on.

 

[CraigGB]

so what now? Video playback has that much difference?
if THAT is a definite fact, we DO have a problem.
don't get me started again!

oh oops that test is still performed on the first phone that he later returned, I worded it badly earlier when I said about the second test. he performed one set of tests with that phone comprised of several tests in that one article. he seems to have returned it and did a bunch of tests with a second phone in the follow up article which provided much closer results which I posted above. he never actually says but it seems his first phone might have had issues

 

[btrach144]

Who cares

Just enjoy the damn phone you bought

You guys are truly, really horrible. Just go live your life a little and actually enjoy the phone.

You two must have a Samsung chip. I feel sorry for the two of you. My TSMC chip rocks!!!!!!!!!!!!!!!!!1

 

[j2ee]

personally I think it's far too early to jump that conclusion, the panic seems to have driven a couple of people insane, and really the evidence seems to show the geekbench battery test is an anomaly, so far we have:

the geekbench battery test which shows a massive difference between the two chips.

all of anandtech's benchmarks minus the geekbench battery test show the two to perform within a few percent of each other, the geekbench battery test of course shows the TSMC chip massively ahead.

tomshardware performed their own suite of benchmarks and also found the two chips within a few percent of each other.

several other sites have found the same with real world usage tests, there was a german one somewhere in one of the threads with a pretty good real world test between the two which in the end showed them to perform nearly identical and battery life was within two percent between the two, the TSMC finished at 38% and the Samsung was at 40%, a difference that could easily be down to manufacturing variance.

several video's running a variety of tasks in a real world setting show the performance to be very close, I've included the most thorough in this post that runs several CPU intensive tasks for a long amount of time, the end result is they are within 1 percent of each other in regards to battery and a minor difference in length taken to perform the tasks, again perfectly normal variations in the silicon lottery. they also run the geekbench battery test and it as usual it shows a massive difference between the two, it's possibly the best evidence I've seen that the geekbench battery test is not just an anomaly but it seems to be flawed, however I have no idea what might be causing it, hoping it's figured out soon.

apologies for the long post, honestly I'm just waiting to see what's going on. I would rather wait until we know the answers before forming a conclusion, anyway, haha it feels like I've recounted the last couple of days just about in regards to chip gate, the video I mentioned is below.

Base on german test, the whole battery different would be around 4%, many people in this forum report they feel around 10% battery different after they exchange the phone. Samsung one definitely is the suck version, the question is just suck by how much.

 

[j2ee]

The following is a CPU-intensive test, other than the Geekbench, without background programs, with minimal influence from other components such as display and wifi.

"The brightness and volume are adjusted to minimum, and to turn off automatically to reduce the influence of the screen. ... The phones are fully charged to 100% and restarted.

"Test 1: Safari Javascript rendering complex animations. ... CPU utilization is running about 60%. After 38 minutes, the Samsung 14nm has 59% charge remaining [but the] TSMC 16nm has 74%."

http://www.computerworld.com/articl...dware/apple-iphone-6s-batterygate-itbwcw.html

Based on the above test result, Samsung iphone consumes 41% of a full charge; TSMC, 26%. So, TSMC is better by

(41-26)/26 = 57.7%

Or, put it another way:

On the same CPU-intensive task, TSMC phone uses 26 units of power; while Samsung consumes 41 units.

26 * 1.577 = 41. Therefore, Samsung uses 57.7% more power.

Tentative conclusion:

Many tests, Geekbench or others, had shown Samsung A9 has far worse power efficiency, under heavy CPU workloads.

However, since components other than A9 consume most power and not everyone runs CPU-intensive programs, the total battery gap on the iphone may be only in the single digits.

Yeah if we just count cpu usage, samsung chip has serious problem, but as I said earlier "you will just drive your expensive sport car in city and get traffic jam anyway", then people make themsleves happy by seeing not using cpu much in real life as arguement.

There is no way to consider a main component energy efficiency is around 30 to 50 percent worse than same model and say QA is ok.

Apple would do anything to prevent recall around 50 percent of new phone, iphone is 70 pecent revenue of apple.

 

[Shrek Chung]

Does anyone see the strange temperature on the Apple logo of Samsung?
Its temperature is cool same as the room.
It seems only samsung has the fan

 

[Kal-037]

Wait, mine came with an AMD chip... should I be worried?


Kal

 

[HEK]

Dont worry there will be another article next week saying the tmsc is better haha. At this point just keep your phones and enjoy it. If your getting poor battery life you may have another issue.

This will keep up till about August next year, then we start all over again. Know what I mean.

 

[kvnt]

Video link:

The above test was on continuous web page viewing, with all components on, the TSMC power advantage drop to only 7.34% or 9%, in the range of single digit percentage I had guessed, because other iphone parts combined consume much more power than A9 alone.

For example, if A9 constitutes only 15% of total power consumption, 15% out of 40% TSMC advantage is only 6%.

 

[evorc]

I was wondering why the other two threads got quiet. So this is where you guys are barking now.

 

[kvnt]

The Geekbench battery results reached 40% in TSMC’s favor, when testing A9 alone, that is, tested with other components (display, wifi, cellular) off.

The above non-Geekbench test of continuous web page viewing, with all components on, more or less, confirms the Geekbench results.

Summary:

40% TSMC advantage without the influences of other parts; dropped to 7-8% on the entire iphone (including the power consumption of all components).

BTW, Apple’s official statement of 2-3% difference may be correct, after all. Continuous web page viewing generates only 7.34% delta. Since very few keeps re-loading web pages, the real Internet browsing may indeed render no more than 2-3% delta.

 

[j2ee]

The Geekbench battery results reached 40% in TSMC’s favor, when testing A9 alone, that is, tested with other components (display, wifi, cellular) off.

The above non-Geekbench test of continuous web page viewing, with all components on, more or less, confirms the Geekbench results.

Summary:

40% TSMC advantage without the influences of other parts; dropped to 7-8% on the entire iphone (including the power consumption of all components).

BTW, Apple’s official statement of 2-3% difference may be correct, after all. Continuous web page viewing generates only 7.34% delta. Since very few keeps re-loading web pages, the real Internet browsing may indeed render no more than 2-3% delta.

Geekbench result is not fake or anti samsung chip, it is just the worst case when we use nothing else but just the cpu.