3 ms·

I had a similar problem with an iPhone 6. The battery page in Settings said the battery was healthy but the phone wasn’t lasting long and was benchmarking at a

I had a similar problem with an iPhone 6. The battery page in Settings said the battery was healthy but the phone wasn’t lasting long and was benchmarking at about half the score it should have. Coconut Battery said the battery was around 65% of its design capacity. After replacing the battery the phone went back to normal, benchmark scores improved. It’s really scummy that the phone was downclocking itself to lighten voltage but not informing me. This was all after the iPhone 6 battery scandal and software updates to show battery health.

derefr 5y ago
If it was after the scandal, and you turned the battery-health protection OS “feature” off, then I find it extremely unlikely for it have been the OS doing that—actively lying about what that toggle does would have caused another, even bigger scandal. Instead, it may have just been a direct, low-level interaction between the battery and the CPU — with the CPU saying “I need more power!” and the battery replying with “she canna handle any more, cap’n, she’s gonna blow!” (Or more literally: the battery being unable to supply the spike voltages the processor needs in order to boost, and so the CPU’s VRMs noticing that the boost “isn’t working” [by getting browned out] and signalling the CPU to stop trying. Or, alternately, the battery heating up when trying to provide that power, due to increased internal resistance, to the point that the heat so produced, triggers thermal throttling in the CPU. Either way, a CPU that couldn’t get its highest frequency-scaling multipliers going for more than a few milliseconds before stalling and dropping back down. Which would tend to feel slower, and lead to lower benchmarks.)
- CoolGuySteve 5y ago
  Yeah, every time I post about this somebody on here comes and tells me the Battery Health monitor can’t possibly be lying and that my lived experience must be wrong. The Battery menu didn’t even present the togggle in the first place because it assessed my battery as healthy. Just like with Rachel here, the Battery menu seems to miss a lot of fucked up batteries. It’s not immaculate. But if the voltage is not being supplied correctly I would consider that to be degraded “Battery Health”.
  - derefr 5y ago
    Please don’t conflate “the causal chain I have mentally-modelled to be responsible for my lived experience” with “my lived experience.” I don’t doubt that you saw what you saw; but that doesn’t mean that you’re suddenly uniquely qualified to judge what particular interaction of hardware and software features could have led to you seeing what you saw. > But if the voltage is not being supplied correctly I would consider that “Battery Health”. “Battery health” as a feature — i.e. the thing Apple implemented as a response after being sued about not having exposed it to users — only tracks what voltage the battery is putting out, compared to what voltage it would be expected to put out given its lifetime number of discharge cycles and current charge. Which is all the previous, hidden algorithm for “battery health” was basing its decisions on, and so is all that got exposed for manual control. (As it happens, it’s exactly the same algorithm that leads to a macOS laptop saying “Battery Needs Servicing.”) This algorithm is meant to detect one thing: whether your battery is like a leaky gas tank, where the amount of voltage * amperage you get out, is less than the amount you put in. Unlike other battery problems that are down to faults in the battery, this problem is an inevitable (100% eventual failure-rate) problem for old, used batteries to suffer, so it’s important to remediate and impractical to offer free replacements for. It’s literally just “wear and tear”, like, say, tyres going bald. If a battery issue is something that only happens when the CPU boosts i.e. if it’s flaky in a “sometimes perfect, but situationally does the wrong thing” way — then that’s not regular wear-and-tear, but is instead an actual fault in the battery. The battery health tracking algorithm has no way of knowing† if your battery is faulty. If it could, it would probably say “this battery is broken, please contact Apple for a free replacement.” Because a faulty battery, of exactly the type you described, is a warrantee-covered problem. † Really, no software can know/predict how a given power source will cope with increased load. This is why we don’t have OS logic in desktop PCs that can convert “a PSU too weak to power your CPU + GPU together at maximum load” into “so they down lock” rather than “your computer spontaneously powers off.”
    - AlisdairO 5y ago
      seems like you could get a reasonable idea by performing a brief workload that should reliably trigger boosting, and see if boosting occurs?

derefr 5y ago

If it was after the scandal, and you turned the battery-health protection OS “feature” off, then I find it extremely unlikely for it have been the OS doing that—actively lying about what that toggle does would have caused another, even bigger scandal. Instead, it may have just been a direct, low-level interaction between the battery and the CPU — with the CPU saying “I need more power!” and the battery replying with “she canna handle any more, cap’n, she’s gonna blow!” (Or more literally: the battery being unable to supply the spike voltages the processor needs in order to boost, and so the CPU’s VRMs noticing that the boost “isn’t working” [by getting browned out] and signalling the CPU to stop trying. Or, alternately, the battery heating up when trying to provide that power, due to increased internal resistance, to the point that the heat so produced, triggers thermal throttling in the CPU. Either way, a CPU that couldn’t get its highest frequency-scaling multipliers going for more than a few milliseconds before stalling and dropping back down. Which would tend to feel slower, and lead to lower benchmarks.)

CoolGuySteve 5y ago
Yeah, every time I post about this somebody on here comes and tells me the Battery Health monitor can’t possibly be lying and that my lived experience must be wrong. The Battery menu didn’t even present the togggle in the first place because it assessed my battery as healthy. Just like with Rachel here, the Battery menu seems to miss a lot of fucked up batteries. It’s not immaculate. But if the voltage is not being supplied correctly I would consider that to be degraded “Battery Health”.
- derefr 5y ago
  Please don’t conflate “the causal chain I have mentally-modelled to be responsible for my lived experience” with “my lived experience.” I don’t doubt that you saw what you saw; but that doesn’t mean that you’re suddenly uniquely qualified to judge what particular interaction of hardware and software features could have led to you seeing what you saw. > But if the voltage is not being supplied correctly I would consider that “Battery Health”. “Battery health” as a feature — i.e. the thing Apple implemented as a response after being sued about not having exposed it to users — only tracks what voltage the battery is putting out, compared to what voltage it would be expected to put out given its lifetime number of discharge cycles and current charge. Which is all the previous, hidden algorithm for “battery health” was basing its decisions on, and so is all that got exposed for manual control. (As it happens, it’s exactly the same algorithm that leads to a macOS laptop saying “Battery Needs Servicing.”) This algorithm is meant to detect one thing: whether your battery is like a leaky gas tank, where the amount of voltage * amperage you get out, is less than the amount you put in. Unlike other battery problems that are down to faults in the battery, this problem is an inevitable (100% eventual failure-rate) problem for old, used batteries to suffer, so it’s important to remediate and impractical to offer free replacements for. It’s literally just “wear and tear”, like, say, tyres going bald. If a battery issue is something that only happens when the CPU boosts i.e. if it’s flaky in a “sometimes perfect, but situationally does the wrong thing” way — then that’s not regular wear-and-tear, but is instead an actual fault in the battery. The battery health tracking algorithm has no way of knowing† if your battery is faulty. If it could, it would probably say “this battery is broken, please contact Apple for a free replacement.” Because a faulty battery, of exactly the type you described, is a warrantee-covered problem. † Really, no software can know/predict how a given power source will cope with increased load. This is why we don’t have OS logic in desktop PCs that can convert “a PSU too weak to power your CPU + GPU together at maximum load” into “so they down lock” rather than “your computer spontaneously powers off.”
  - AlisdairO 5y ago
    seems like you could get a reasonable idea by performing a brief workload that should reliably trigger boosting, and see if boosting occurs?