Apple slows your iPhone down "to conserve battery"

[james.miller]

drunkenmaster knows what he's talking about.


What I don't understand is why don't they simply accomplish this using screen brightness.

Android has this feature when you put the screen above 80% it warns you that it will be using too much battery.

They could have simply limited the screen brightness. Less photons being ejected from the screen does certainly equate to less energy usage. This nonsense with limiting CPU performance which means it has to work longer to perform the same task is just nonsense.

What and limit the screen brightness to 50% instead? i dont see that being any better.

 

[asim18]

What and limit the screen brightness to 50% instead? i dont see that being any better.

No, simply let the user know that their battery can't maintain enough voltage and simply recommend that brightness is kept as low as possible.

Halving the amount of light output is a definite reduction in power use. Halving CPU power is just pointless it doesn't half the amount of calculations the CPU has to accomplish.

 

[asim18]

Also, another reason halving CPU performance is pointless is because the actual software is becoming less optimised and more bloated anyway.

 

[Threepwood]

I wish my iphone2 would die.

Had it since release, and it has been on every single day.

 

[james.miller]

No, simply let the user know that their battery can't maintain enough voltage and simply recommend that brightness is kept as low as possible.

Halving the amount of light output is a definite reduction in power use. Halving CPU power is just pointless it doesn't half the amount of calculations the CPU has to accomplish.

No it isnt pointless as i explained. Not that either solution is acceptable, but halving the cpu speed changes the usage of the big and little cores and the smaller cores ARE more efficient even if they take longer to do something. that's the whole reason they exist in the Big.little architecture.

 

[Six6siX]

A PC that has throttled itself to lower power modes will also likely be turning off things like hard drives that aren't in use and haven't been accessed for X minutes.

In the micro server instance I mention, it specifically only affects the CPU which is why I used it as an example. The low power mode there guarantees the maximum power draw over a given time period which is always lower than that of the 'normal' modes where CPU speed (and power draw) may be variable up to a higher level when needed. Given a finite energy source (a battery, which may be degraded) and the variability of use cases, this method makes it a certainty that in low power mode, the uptime will be longer for a worst case scenario. It will be will be slower, of course, but it will last longer because the overall power consumption over time is less. This is what this is all about - maintaining a usable length of time for a mobile device with a degraded battery condition.

No it isnt pointless as i explained. Not that either solution is acceptable, but halving the cpu speed changes the usage of the big and little cores and the smaller cores ARE more efficient even if they take longer to do something. that's the whole reason they exist in the Big.little architecture.

Very good point, I'd forgotten about the architecture - makes scaling performance/power usage much more effective. I suppose core shutdown on desktop architectures would be the closest thing.

 

[Bear]

No it isnt pointless as i explained. Not that either solution is acceptable, but halving the cpu speed changes the usage of the big and little cores and the smaller cores ARE more efficient even if they take longer to do something. that's the whole reason they exist in the Big.little architecture.

If you take longer to do something then the phone will be churning for longer with the screen on for longer. Seeing as the screen is one of the more power hungry elements of a phone the savings on the battery will be little to nothing. If the phone was churning away with the screen off then yes there would be significant power savings but phones generally do very little when left alone.

 

[james.miller]

If you take longer to do something then the phone will be churning for longer with the screen on for longer. Seeing as the screen is one of the more power hungry elements of a phone the savings on the battery will be little to nothing.

Next to nothing? are you saying you know more about this than ARM does?

 

[drunkenmaster]

Intel are limited by the fact that they only make hardware. Apple design everything from the CPU to the operating system to the core apps used on their phones. They can do the kind of joined-up thinking that’s impossible for Intel.

Not all tasks require a fixed amount of calculations. For example, PC games offer a range of graphical settings. Apple have the ability to drop or simplify tasks for slower CPUs.

This is completely beside the point. If you're doing 7000 calculations or 70000 calculations, it's more efficient to turn the cpu up, get them done and power down. A phone over a day will power up and day thousands of times to do different work, mostly very simple stuff. Most people don't run benchmarks or games all day long, but basic things, most of which a more powerful cpu can't really do differently, just faster. Slower cpu = longer on time = worse battery performance. This has been true for a very long time and every single mobile cpu maker designs to be faster and on for less time and have been designing this way for 5+ years. The calculations are irrelevant, it's about how fast you get back to idle power levels, faster the better. Turning the CPU down is inefficient and absolutely worse for battery, optimising software is a great way to improve battery life.... but iOS updates and software updates are largely blamed for reducing battery performance on Apple phones, not increasing it.

 

[drunkenmaster]

I'm guessing the reason Apple seem to be having this problem when it doesn't seem some of the other phone manufacturers are, will be down to the choice of battery design and power regulator - they mention the battery not being able to provide enough power when it's aged for full load resulting in shut downs, that suggests to me that the battery design is such that it isn't able to provide the current the chip that regulates the power needs*, or that chip isn't able to handle the situation as well as it should.



*The whole idea of a power regulator chip is to keep a steady output voltage and current regardless of the input voltage within the design range of that specific chip.

Unless Apple has done something absolutely crazy, truly crazy with the voltage regulators, then there is no way even a heavily degraded battery is having trouble providing the right power for the voltage regulators to provide likely at most around 1v to the cpu from a 3.8v or more battery. If they have made voltage regulation so spectacularly bad that it can't cope with battery degradation at all, then people shouldn't be buying the devices because they simply aren't fit for purpose and certainly not worth the money they charge.

It's an excuse, nothing more or less. It sounds technical enough to make enough people believe it's true and really it was the only excuse they even attempt to use.

 

[drunkenmaster]

No it isnt pointless as i explained. Not that either solution is acceptable, but halving the cpu speed changes the usage of the big and little cores and the smaller cores ARE more efficient even if they take longer to do something. that's the whole reason they exist in the Big.little architecture.

Big.little is more about loads and types of workloads, they aren't all just uniquely more efficient. They are more efficient in scenarios that the load doesn't outweigh the performance. IE for basic/low workload that can fit into a 1.2Ghz clock speed where running at 2Ghz wouldn't improve performance but would waste clocks then it's more efficient to run on the lower clocked cores. If however you have a heavy workload then it becomes more efficient to put that load onto the higher performance core so it completes more quickly. That is pretty much the point of big.little, splitting workloads to the appropriate core.

In reality in day to day use there are some times you'll need slightly higher performance, other times you need lower performance. Having lower performance because they force you onto the lower performance cores is, well, it's selling you a big.little chip but only allowing you access to the little cores after a certain amount of time, if that isn't advertised it's entirely and utterly inappropriate.

More importantly, before the battery is reduced all the workloads that are already more efficient on the little cores should already be on the little cores and all the workloads on the heavier cores should be more efficient on the heavier cores, so they are pushing a heavier workload to a less efficient core which makes no sense.

More importantly, people are reporting this more on older phones which didn't have a big.little setup, only the last couple phones have big.little. But again the idea of selling you a phone that achieves a certain performance level, and then removing that performance level at a later date is utterly absurd and more importantly, not standard in the industry. Do laptops reduce cpu power as the battery degrades, do any other phones, to mp3 players, do dvd players, do basically any battery powered devices decide to half available performance when the battery gets worse, nope. It's standard in the industry for users to know batteries run out and need replacing but entirely non standard for users to expect their devices to artificially limit their speed just because the battery holds less charge.

I'd be entirely fine with Apple doing this if they told people upfront they would do this, but they don't. You can't subvert industry norms completely without telling people and then expect people to think it's okay.

 

[Things change I've changed]

This is design flaw plain and simple and Apple tried to cover it up by lowering the performance of the phones without peoples knowledge, how can it be seen any other way?

From Sky,

"In a statement, Apple said that, as they age, lithium-ion batteries used in its phones become less able to provide the top levels of electrical current needed.

The problems with peak current draws especially occur when batteries are cold or low on charge, which can result in the device unexpectedly shutting down to protect its electronic components, Apple said."

My Z3 is 3 years old now, I don't have this problem and I still get a good days worth of juice out of it.

 

[Tunney]

Slower cpu = longer on time = worse battery performance. This has been true for a very long time and every single mobile cpu maker designs to be faster and on for less time and have been designing this way for 5+ years.

So why does modern ARM silicon contains a mix of high and low compute performance cores? Why not have fewer but all high compute performance cores?

It’s because most tasks don’t max out the CPU and so spinning up a lower power core is more efficient. Remember that processing tasks are rarely just a series of computations. For example, reading something from disk will be limited by the disk access speed.

 

[james.miller]

More importantly, people are reporting this more on older phones which didn't have a big.little setup, only the last couple phones have big.little.

The oldest A10 equipped products will be 15 months old now. I think it's a fair bet that people will start having problems right about...now. Which is longer than the battery on my note 4 lasted.

 

[Bear]

Next to nothing? are you saying you know more about this than ARM does?

No of course not, don’t be so stupid. We are talking about the slowing down of the performance of the phone to preserve battery power. I was saying if you have a slower processor which meant you have longer on screen time as a result or you had a full speed processor and as a result finished your task quicker so less on screen time then the resultant power saving would probably be little to nothing seeing as the screen is one of the most power hungry elements of the phone aside from the processor.

 

[james.miller]

That's not how Big.little works and we arent talking about single tasks that require the screen to be on for significant amounts of time. Keep your insults to yourself, Bear.

 

[Bear]

Where in any of my posts did I say anything about little.big.? You were the one being insulting with your ridiculous comment appealing to authority.

Any usage of the phone requires screen time, the phone does very little otherwise.

 

[Diddums]

MERRY CHRISTMAS EVERYONE

 

[james.miller]

Where in any of my posts did I say anything about little.big.?

*I* was talking about it, you responded to *me*.

You were the one being insulting with your ridiculous comment appealing to authority.

Ridiculous comments? Nope, lost me. Not sure what your issue is but it not my fault.

Any usage of the phone requires screen time, the phone does very little otherwise.

The point was anything that taxes the cpu to the extent that the screen needs to be on for addition seconds / minutes isnt going to save anything by running on slower cores instead. That isnt the point of Big.Little as i explained; Big.Little is about effective management of tasks - the right hardware for the job. Decoding an MP3 is a perfect example. You dont need to use the the power hungry cores to do that. The efficient cores might take a fraction longer to do the job, but they'll do it more efficiently. Limiting the cpu frequency could shift some of the tasks that *just* would have gone to the bigger cores, to the smaller ones. Sometimes it'll still be more efficient, some times it wont. Anything that taxes the big cores though, probably wont be more efficient on the smaller cores no matter what the task is. Lower clockspeeds should mean lower voltage as well.

I know, ridiculous right ??

 

[Nasher]

Lol Apple...

They only use excuses like this because most of their customers don't know better. The battery life is never an issue since the user can simply pop the back off and replace it. Just like you can with an SD card. Oh wait...

Whats next? A glass screen on a device which is likely to get dropped at some point? Only a moron...