TSMC's A9 Chip Outperforming Samsung's in Early iPhone 6s Battery Benchmarks

[mdlooker]

So Apple got hit with major backlogs for previous models. To avoid that and better meet the demand, they go to two chip manufacturers. So now the cut of the double-edged sword... though both phones may meet the specs provided and listed by Apple, there still is the real desire to have the better performer.

Guess Apple can't win for losing.

 

[mjcxp]

How bad is bad? I also have the 6S+, but TSMC.

During the year I had my 6+ I cannot remember ever reaching 0% during a day. I have reached 0% multiple times already with my 6S+. I still think the 6S+ is better overall because the A9 SoC is incredibly fast, which makes up for any of the battery life shortcomings.

 

[scottytwo]

One phone has working sim installed.
One does not.

Not controlled enough of a test.
Reproduce, verify, repeat!

 

[matthewadams]

Waiting for some tests on iOS 9.1 due to vastly different kernel versions than 9.0.x ( https://www.theiphonewiki.com/wiki/Kernel#Version_List )

Wondering if there were any under-the-hood improvements addressing this..

 

[wes-]

Waiting for some tests on iOS 9.1 due to vastly different kernel versions than 9.0.x ( https://www.theiphonewiki.com/wiki/Kernel#Version_List )

Wondering if there were any under-the-hood improvements addressing this..

Since iOS9.1b4 my battery life (iPhone6s/N71AP) has been improved a lot it seems.

 

[Merkie]

We know Apple buys chips form 2 different providers, but there could be more providers for batteries too.
In my opinion, comparing battery life without considering if there are different brand of batteries is futile.

It depends on what you're trying to prove. If you want to prove that there is a correlation between the source of the chip and battery life, all you have to do is run a sufficient number of battery tests in a scientific way. If you want to prove that the chip is to blame for this, then indeed you are right.

 

[ds2000]

Seriously, maybe his 6s with the Samsung chip just has a bad battery. 2hrs difference seems way to obvious. I really don't think Apple is that stupid.

Doesn't it just. I'm so glad end users do testing Apple wouldn't have thought of

 

[scaramoosh]

Mine is using the Samsung one, it only takes 5 mins to run down the battery, it grew legs, shot up a School somewhere in the US and is producing so much heat that NASA are now sending it in a rocket to Mars to try and restart it's core.

Although I did see this that kinda debunks my claims..

 

[MarcusCarpenter]

TSMC 64gb gold 6s
MLC nand 64gb gold 6
Sweet

 

[JustMartin]

Cue all the 'I have the Samsung chip and my battery life sucks' posts now...

No, I'm waiting for the "I had the TMSC chip and made them change it to the Samsung model and now I'm going to go and make them change it again" posts

 

[Substance90]

People without a background in statistics should not use the term significant so causally. With an N of 1 you can't claim a statistically significant difference. There may be real differences, but anecdotal crap like this just fuels clicks and wild irresponsible posting.

That!

 

[astroboy888]

It appears that a "burn test" is being performed - Stress testing the CPU until it runs out of battery. It is certainly not representative of real world usage, but it is a data point in comparing the two CPUs. This could very well be an isolated incidence, but a demonstrable 2 hour difference in the "burn" test clearly shows a defect in the iPhone6s with Samsung produced A9. It could be due to the battery, who knows. The point it is, it is defective and should probably be replaced by Apple.

Not sure we all know the details quite yet, but there is a lot of data suggesting at least this particular battery benchmark app is showing a significant difference between the Samsung and TSMC. I think that enough folks have done it now to say that there is definitely a statistical difference (although maybe not with a "scientific" approach, or whatever the **** that is to mean).

What this leaves us with is this:

• Results don't reflect real-world usage.
  • Although both chips are A9, it is possible that the differences combined with the methods used by this battery benchmark are leading to these results, and that real-world usage doesn't follow.
• Results are reasonable and real-world usage shows TSMC gets better battery life
  • Samsung iPhone 6s may still meet Apple's specifications, just the TSMC iPhone 6s exceed it.
  • Samsung iPhone 6s may not meet Apple's specifications. When (if) that conclusion is reached I'll be ready with my pitch fork (since I own the Samsung one).

 

[astroboy888]

There is no such thing. All A8s are made by TSMC.

I have an iPhone 6 with an A8 Samsung chip. Should I return it?

 

[Gazous]

I have a Samsung iPhone 6S one and I am just waiting for more serious testing...

 

[astroboy888]

There are few possibilities. 1. bad battery. 2. Bug in low power circuitry. (Sadly it happens quite often). 3. Problem with semiconductor manufacturing. Too much variation in certain key process step, which causes power drain to manifest subjecting the CPU to certain temperature, voltage, and functional mode. These kind of bugs are really hard to diagnose. But it does happen.

Am I missing something here? If the Samsung based device is running about 22% shorter can that even be explained by difference in chips?

If the A9 accounted for half the power in the phone, doesn't that mean the Samsung is like 50% less efficient? That sounds way outside of quality control limits... The only thing I can imagine causing that is if the Samsung had a bug in its idle mode, which Apple would have had to have known about.

 

[SoGood]

At the end of the day, if Apple quoted the lower performance spec of the two, then they have satisfied their responsibilities to the consumer. It's a bonus which chip one received. Until further verification, we are yet to know which chip is the 'better' of the two. Waiting...

 

[scaramoosh]

There are few possibilities. 1. bad battery. 2. Bug in low power circuitry. (Sadly it happens quite often). 3. Problem with semiconductor manufacturing. Too much variation in certain key process step, which causes power drain to manifest subjecting the CPU to certain temperature, voltage, and functional mode. These kind of bugs are really hard to diagnose. But it does happen.

Yeah I mean my 5s had terrible battery, it was draining overnight, listening to a podcast for 2 hours would cause it to lose a third of it's battery. My Mum had a 5 and she would have to charge three times a day....

Sometimes you get devices with a faulty battery, luckily even though mine tested fine, they gave me a replacement any ways and the second 5s I had was fine and I was getting 2 days out of a single charge.

It's more than likely that they have a dodgy battery.

 

[TripHop]

That's quite a big difference. Glad I've got the TSMC chip

Me too. Relieved to find mine's the TSMC A9 also.

 

[astroboy888]

Haha no way, the only way you can cheat a benchmark is by modifying the OS or software. You can't do that inside a chip. The only people who have access to OS and OS level API is Apple.

And we should see if there is CPU throttling or not, 'cause that affects the devices as well. If there would be cpu throttle, it gets more battery run time but lower performance. If there would not be cpu throttle then it gets higher performance but lower battery score. So TSMC version is 16nm and it could have been more cpu throtlle than the 14nm Samsung manufactured A9 chip.

Kind of like a VW emission test.

 

[astroboy888]

TSMC can't figure out marketing if it came and bite's its ass. The largest contract semiconductor manufacturing company in the world for 30 years and no one hears about it. LOL!

TSMC marketing has been learning from Samsung and is behind this '-gate'?

 

[SoGood]

TSMC can't figure out marketing if it came and bite's its ass. The largest contract semiconductor manufacturing company in the world for 30 years and no one hears about it. LOL!

Until they make chips that we consumers can plug directly into my brain without a middleman, they have no real need to advertise to us. Then there's that 'Intel Inside' campaign...

 

[astroboy888]

No smaller is not always less power I'll let someone smarter explain why

Stolen Tech? Please. Share your source. "Mix Process?" The 14nm FinFET SoC APL0898 is just that. The APL1022 TSMC 16nm is a stacked FinFET, the 0898 a Die-Shrink. The 0898 should be more efficient and cooler. This however does not seem to be the case. So, I find it interesting.

Next time don't hold back on your TSMC bias.

Here ya go. Samsung director responsible for 14nm, former TSMC employee working on 15nm, convicted of trade-secret and internal plan theft which was found to directly benefit Samsung's 14nm process.

Next step is probably international courts.

http://arstechnica.com/tech-policy/...secrets-to-samsung-taiwan-supreme-court-says/

 

[hello09381]

I have the Samsung model and my battery life has been consistently mediocre since the day I got it.

What a crock this is, I avoid Samsung phones and I still get screwed over by the garbage they put out.

 

[Palladium SG]

I have the Samsung model and my battery life has been consistently mediocre since the day I got it.

What a crock this is, I avoid Samsung phones and I still get screwed over by the garbage they put out.

Assuming that Samsung A9s are in general worse performers I can understand why Apple is trying to dump them so freaking hard.

 

[SoGood]

I wonder how's the split b/n TSMC and Samsung's A9s amongst iPhone of different capacities? I have a 128GB 6s and it came with a Samsung. Assuming the Samsung has higher performance and power consumption, would Apple preferentially use the Samsung A9 in the higher capacity models to churn through presumed greater data load? While the TSMC are used in the 16GB models.