Apple blames iOS 17 bug for overheating iPhone 15 woes Apple has warned that a bug in the iOS 17 software powering its latest iPhone 15 Pro model is causing smartphones to overheat. Some folks who purchased the new device complained about their devices becoming uncomfortably hot during normal use. The iGiant has now said that’s not unexpected, especially when setting up the iPhone …
That's obviously not the cause, because the phones with stainless steel frames aren't dissipating heat through the frame. My 14 Pro Max barely gets above ambient, but when it does I feel it in a certain place in the back glass (where the SoC is located, no doubt) not on the stainless steel edge.
In both the 14 and the 15 the inside portion of the frame is aluminum.
It will be interesting to see if the software update helps, though I still think it is TSMC's N3B process that is responsible. Having yields that poor (estimated at 55-60%) would lead to a much wider range of required voltage to achieve the target clock speed. That would account for the observed behavior where some 15 Pro/Max phones are claimed to get hot, but others with the same phones claim theirs has no problem (but I guess a software bug that hit randomly would show the same...we'll know soon enough)
55~60% is actually a pretty good yield for a process at this very early stage of production. But, processes don’t have heat problems, they have density limits- if the designer tries to pack high-power sections of the IC too close together, then you get heat problems. Maybe TSMC fell short of their promised density, or maybe Apple has been too ambitious, overspent the increased capability of this process-shrink and then needed to live on the ragged edge just to fit everything in.
My stab in the dark is that Apple’s new GPU cores are at fault - the reported heat issues all have video-heavy tasks as a common factor. That could be software (drivers pushing the GPU beyond its specified safe envelope) or it could be hardware (hardware not living up to the agreed thermal parameters), but really the only way to fix it is software throttling. If it is hardware, it must be pretty dispiriting for the SW Eng teams to once again be publicly blamed for a problem that isn’t of their making; but as consolation to them, I don’t think the general public really believes that this is due to a software problem.
Any of that will result in lots of Class Action lawsuits. We saw that with the previous case where slowing thing down to prolong the battery life got a lot of lawyers all hot under the collar and Apple had to pay out LOADSAMONEY over that.
As for blaming the engineering teams, I would put it more down to differences between pre-production and production hardware.
Changing throttling behavior wouldn't justify class action awards, all phones throttle and no doubt software updates change that behavior over time. If performance went down by double digits then maybe there's an argument, but that's hardly likely. Even if there was some legal basis it would only cover phones that had already shipped. Anyone who bought one after the update would have no leg to stand on, since the performance of their device never changed from the day they bought it.
55-60% is terrible by TSMC standards. TSMC has between 80% and 90% when they began mass production of N5/N7/N10 so this N3B yield is a serious departure, see https://images.anandtech.com/doci/16028/Manufacturing%20Excellence.mkv_snapshot_02.11_%5B2020.08.25_14.16.22%5D.jpg
If the GPU turns out to be at fault then I would guess that Apple is correct that it can be fixed with a software update. The A17 GPU is the first ground up rework since they started making their own GPU, and new GPUs (from all vendors) often have teething problems for the drivers.
If it was only heating up in games heavily exercising 3D I might believe it is a hardware issue, but they are pointing at stuff like Instagram as one of the problematic apps and that's simple stuff that barely exercises a modern GPU. So perhaps there's a combination of Instagram not following the iOS API guidelines and iOS 17 drivers doing things in a different way on the A17's new GPU hardware that creates problems for code not following the guidelines. That would explain what Apple has said they did - notified developers of certain popular problematic apps (for a more immediate fix for that app only until Apple's fix is ready) and a bit later an iOS update that hopefully corrects the issue in all the apps doing it wrong even the (probably many) that Apple hadn't notified.
I'm withholding judgement on any of this until the software bugs are settled down: as this video shows, the Instagram bug can be seen decreasing battery by 1% per minute!
https://www.youtube.com/watch?v=P6X2ZIkYFsQ
After that, I'll re-assess whether it's chip or casing....
Thanks for the link, that's quite interesting. It sure does look like what you'd expect with driver issues for the new GPU, a hardware fault could hardly drain battery that fast when it is just sitting on a static screen. It would almost have to be something causing it to re-render the display at 120 fps. Or maybe Instagram is downloading and caching a bunch of content down below the screen for smooth scrolling, and it is all getting prerendered instead of a limited amount? Because it sure is working hard for sitting still lol
Maybe the driver for the old GPU only rendered a limited amount of off screen data, just doing enough for smooth scrolling (unless you scroll so fast you can barely read) but the new driver doesn't have such a limit, or the limit is way too generous. The Instagram app could have code for "render this entire buffer" and be downloading a lot of stuff to support the doomscrolling a lot of social media users do, and that was fine with the old driver since it had a reasonable limit. However with the new driver perhaps it is working its ass off instead of only staying a little ahead of the end user's scrolling.
That would be not necessarily an Instagram "bug", just an expectation that the OS would intelligently limit rendering of off screen data as it had in the past - but the quick way to fix it today would be getting Instagram to update it as Apple apparently requested they do, since Apple's iOS test process take a bit more time but would provide a permanent fix for all apps that act similarly. Of course even if this is an issue it doesn't mean it is the ONLY issue, with a brand new GPU with a brand new driver this might not be the only "unexpected behavior" causing excessive power draw.
Fuck knows, but I earned mine...it enrages me that people like this get that title willy nilly...
I work on a soft drinks production line, and lieu of a proper payrise they promoted me from Junior Lid Turner to Top Analyst, I used to feel quite important until I realised anyone can be a "Top Analyst".
People like that just cheapen the title.
This post has been deleted by its author
Surely if the new frame couldn't dissipate the heat properly then the phone would feel LESS cold to the touch (even if its innards were being grilled)?
The way I see it the frame is dissipating all the heat properly, the problem is that too much of it is being generated in the first place (definitely not by the frame itself).
