It seems the new iPhone 15 Pro is having overheating issues, and while I normally don’t really care and don’t mention this sort of nonsense, I found Apple’s response to the issue… Peculiar.
Furthermore, Apple tells 9to5Mac that recent updates to certain third-party apps are causing them to overload the system. The company says it’s working directly with those developers to fix the issues. According to Apple, some of the apps overloading the iPhone CPU and causing devices to overheat are Asphalt 9, Instagram, and Uber. Instagram issued a fix for the problem on September 27, Apple says.
Apple designs and builds the SoC, the thermal system, the outer casing, the operating system, the APIs, and is the gatekeeper for every application that runs on an iPhone – and yet the company still blames third party developers? How is it even possible that any of these applications can cause unexpected overheating in the first place, and how, if the App Store review process is put in place to protect users, did nobody at Apple catch this during the review process? If they can’t even detect and stop applications that can physically damage your iPhone, how on earth can anyone trust them to stop malware, spyware, and other crapware?
I can’t believe people still fall for this.
This is the “you’re holding it wrong” defense. Haha. it’s apple’s responsibility alone to keep devices safe from “overheating” and it’s disappointing to hear them play the blame game.
This article made me curious what exactly was meant by “overheating”, apparently their devices are reaching 116F/47C.
https://9to5mac.com/2023/09/26/iphone-15-overheating/
While such temps are unexpected for a mobile device, for all I know 116 is within technical specs for the hardware and iOS is following its programming to enforce those specs. This makes me wonder if something’s actually wrong here, or if apple processors are simply using more wattage than a mobile form factor can comfortably handle? If apple’s fix reduces performance, then I think we have our answer.
That’s… not how it works. If I assemble a computer and it cannot handle the heat load of e.g. rendering in Blender and gaming at the same time, there is only one person to blame for this, and it isn’t the developers of Blender or the game I am playing.
Properly designed system should be able to run at 100% load all day without throttling. Anything less is unacceptable. Because once the designers are allowed to get away with blaming the software for taxing the machine too much, they’ll start doing stuff like incorporating under-spec power supplies and then claiming the same when the system shuts down. “Your software is too rough! See? If you do nothing but watch movies and write emails all day, it works perfectly fine!”
Note that this explicitly does not include overclocking, which I can assure you that Iphone developers are not doing. Once someone starts overclocking a system, all guarantees are void.
But all phones “overclock”.
Basically at rest there is a certain thermal envelope. When the phone is cool and idle, the processor can work at high capacity for a while (until it gets hot).
Obviously the CPU itself, but also the battery and the screen light will heat up the chassis. This should in theory cause graceful reduction in load, providing a balance between comfort and performance.
(There is also a “gaming” mode which adjust some settings, but I don’t know much about the details: https://9to5mac.com/2023/07/25/game-mode-macos-ios-17/. As far as I know, in general dedicated gamers use coolers, etc to increase the high speed durations: https://www.razer.com/mobile-accessories/razer-phone-cooler-chroma)
The problem is, in this case, the OS seems to have a bug, where some applications trigger a high power usage that does not go down as expected. There are more open questions of course…
sukru,
You make good points. Desktop and laptop computers that throttle below CPU specs are frowned upon. But when it comes to things like phones that have no active cooling, they’re just bad at sustaining heavy loads. It’s much more likely that they will have to throttle in order to stay cool. The physics is pretty easy to work out: how many watts can the phone comfortably dissipate versus how many watts does the phone use at full load (cpu/gpu/battery/voltage regulators/wifi/cellular/screen/peripherals/etc). If the watts used is higher than the watts dissipated, then clearly it’s only a matter of time before the phone must be throttled.
A good way to detect throttling is to run a stress test/benchmark twice in a row. If subsequent runs are consistently slower than the first, it’s a telltale sign of throttling.
I would agree with kbd if the throttling caused the device performance to drop below advertised performance, that would be a problem of false advertising. But if the throttling is genuinely reflected in all the performance claims then it is fair IMHO. Still there are a number of ways a manufacturer can try to “cheat”: turning off peripherals, dimming screens, chilling the phone, tuning the phone reach hotter temps, cherry picking tests, running unreasonably short benchmarks, etc. To mitigate these risks it is important we have 3rd party benchmarks and not be so reliant on manufacturers bench-marking their own products.
Alfman,
Yes, there should be some common standards to measuring the “turbo” or “underclock” (whichever way you look at it) especially for benchmarks (this was one at +4C in a air conditioned environment with 25% brightness and device connected to a power source, etc)
Especially at a mobile device, however, it is very difficult to know how much power you can dissipate. As you mentioned everything from the users’s choices (brightness, …), and internal state (battery temperature, …) to the outside environment (ambient temperature, …) will have an effect. (Many in each category).
But I agree, at the end of the day, the user should not feel “cheated”, but rather they have received a good value for their purchase.
sukru,
This reminds me of the difficulty of comparing solar panel claims, which I’ve been researching lately.
The industry is supposed to have standards, like 1000watts/m^2 of sun power hitting the panels, but manufacturing specs are all over the place and many users get significantly less than advertised. Even different brands with the same specs have differing output levels.
Here on the east cost, on a clear summer day I was getting about 70% of the panel’s rated power. I was kind of disappointed before learning this was pretty normal for my area depending on the manufacturer.
Even in Florida, which is closer to the equator than most other states in the US, can expect below average solar power. due to local conditions.
https://solcast.com/solar-radiation-map/united-states/
/off topic
What is ” 100% load” Exactly? Modern CPU’s even ones in phones have a range of speeds they operated depending on demand and thermals. They’re designed to speed up when they can and slow down when they can’t handle the heat. Throttling isn’t an error, its a feature.
You did leave out this part, which shows up right before the quote you pulled:
Apple also says that it’s identified a bug in iOS 17 that makes the overheating problem worse for iPhone 15 and iPhone 15 Pro users. This problem will be “addressed in a software update,” the company says.
So it sounds like they’ve found a bug, are asking app developers to issue updates to their apps, and also issuing a patch.
The fact phone SoCs are allowed to pull so much wattage as to make the phone hot to the touch shows how absurd modern smartphone design has become. This massive power draw often necessitates larger batteries, which inevitably means substantially heavier phones. We are talking about 220gram phones as if this is somehow normal (it’s not, that almost a quarter-kilogramme or half a pound). And despite those phones carrying a battery more befitting to a power bank than a phone (4500mah), they barely make it through the day.
Meanwhile, my Nexus 5 weighs a much more reasonable 130grams and could make it through the day just fine before the piles of JavaScript that is the average website today made it so it takes some effort on the part of my Nexus 5 to render the web. JavaScript is the cancer of the modern web and I hate it. At least Flash was optional (mostly).
Not in defense of Apple, But 2 things here. ..
This temperature report seems well within operating temperature for most units. The second, being heat is commonplace while charging. That’s physics. Just take a look at the apparent battery placement between the two pictures. If the Pro Max was too hot to handle, than 115 degrees must be way too low, which bungs an issue with infrared tech accuracy when tracking heat o the equipment/SW utilized, also if the case was on the unit when tracking the temperature.
(Since I couldn’t Edit my comment anymore) It would be plausible if that 115 degrees reading was Celsius. But I assumed Fahrenheit.
The thermal pictures I linked to show “46.7C” at the surface. The thermal images and reports suggest it was based on activity and not charging.
I think that it may well be normal for this phone too. Apple’s response does not make clear this temperature is unexpected for this hardware. They said they would not, but it will be interesting to see if apple reduces performance. They blamed applications, however even if applications are guilty of being poorly optimized, any “overheating” is still 100% apple’s fault and not the applications. It’s fine to optimize applications, but they are technically deflecting here. They also blamed another unspecified IOS bug which may be related to the phone’s background functions.
Logically if they only fix the problem by reducing the load in highly demanding applications & background functions, then that changes nothing with regards to the temperatures the phone can reach under high load, it will just happen in different demanding applications. If they fix it with tighter throttling curves, that could permanently fix the overheating but obviously apps will loose performance.
If they are somehow able to fix this without reducing performance, the only thing that comes to mind is the CPU being powered with higher voltage than required. If that’s the case, I can understand why apple doesn’t want to admit their mistake, but reducing voltage will certainly use less power. For modders, overvolting and undervolting are very useful for overclocking and underclocking CPU/GPU/memory. The goal is to use the lowest voltage that won’t crash the system.