5 ms·
> Knowing the battery size (which also limits the total peak power draw from the battery) and software system in play, Apple could have (and I would argue shoul
by JimDabell 9y ago
> Knowing the battery size (which also limits the total peak power draw from the battery) and software system in play, Apple could have (and I would argue should have) designed a processor that would provide consistent performance throughout the life of the phone.
> Yes, this would result in lower peak performance for iPhone devices.
Aren't they essentially arguing that Apple should enable the throttling on day one instead of waiting until the battery degrades? How does that benefit anyone?
- sandworm101 9y agoNo. It's about building a chip that isnt going to crash when the voltage drops for a picosecond, allowing throttling to happen only as needed without fear of crash.
- ghshephard 9y agoAre there processors out there that can handle voltage drops without crashing?
- ComputerGuru 9y agoPretty much all processors start off that way. The clock speed needs to drop to compensate, though. Unless you actually put some thought into your hardware designs because you're making phones that have a life expectancy of greater than a mere 12 months that will be used around the world for years to come... in which case you can easily reference an electrical engineering textbook and design a circuit that can buffer enough power and supply sufficient voltage/current to meet your expected max sustained power draw with some very simple and cheap components. But simple frequency scaling based off of _dynamically detected_ available power. There's no need to slow down a CPU based off the age of the battery so you can "guess" what the max speed you can safely run at without inducing a brownout is (and even less of a reason to do so in discrete steps instead of linearly scaling the performance, except if you're trying to do it discreetly) - all you have to do is monitor available power and scale the CPU clock rate accordingly in realtime. But if your circuit is poorly designed with no forethought as to what you'd do if the maximum rated power of the battery wasn't available, I suppose that's not so easy to retrofit.
- sliverstorm 9y agoModern x86 chips have built in support for voltage drops, as part of margin recovery. As the chip changes activity profile increasing current draw (e.g. turning on all cores for multithreaded code) the voltage will drop until the regulators catch up. "Back in the day" you used margin to accommodate this. Today, circuitry detects these events on the fly and slows down the chip for a few microseconds waiting for the voltage to recover. This allows you to remove margin and operate closer to the limit.
- tmd83 9y agoEssentially they designed a processor that cannot handle the normal variance of real life usage and so did a software fix that's very expensive to the user but very cheap for them. I am wondering is there a reason they can't be sued to oblivion for this?
- QAPereo 9y agoSued? Yes. To oblivion? No... not enough harm done, and Apple has good lawyers and deep pockets.
- alkonaut 9y agoIt's a tradeoff. Building a chip with more margin costs perf and/or power too. So it's a game you can't win.
- sandworm101 9y agoNo. You can build circuit to accommodate small voltage fluctuations without crashing things, defeating the need to drop the clock speed 24/7. Lots of other chips do this. Apple dropping clock speed as phones age is a very convenient excuse for an anti-feature to sell new phones.
- deckar01 9y agoYes, and I think this is a common strategy in the CPU manufacturing world. The companies developing and testing software for these devices probably don't realize that the performance of their software will start to degrade in 18 months.
- gm-conspiracy 9y agoI think this is a really important point. Also, consider the effect it has on companies that have contracted that development out, and a year later, their app "sucks", without any code changes.
- sliverstorm 9y agoIf you design a chip with the whole lifecycle of the battery in mind, you can achieve better average performance for the life of the chip. Pulling numbers out of the air, the argument is the chip was designed for maximum performance with a fresh battery, 100% performance. But later it's throttled to 80%. Or, Apple could have designed the chip for a steady 95% performance throughout the life of the phone.
- greedo 9y agoI would prefer a phone that lost performance over time than one that was gimped at the start. I think the optics aside, most consumers would prefer that as well.
- syrrim 9y agoDepends on the customer. If you're planning on keeping it for five years, you'd like it to last the time. If you're planning on replacing it when the new one comes out next year, you'd rather it got maximal performance during that year. What it comes down to is which customer apple wants to cater to which is fairly obviously the latter.
- greedo 9y agoKeeping a phone for five years isn't something I'd want considering the rate of performance improvement. A five year old iPhone would be an iPhone 5. The improvements in CPU/GPU since then are fairly astounding. Going from an A6 dual-core at 1.3GHz to an A11 six-core at 2.39GHz is just nuts. One is 32nm, the other 10nm. Smartphones are the one area in computing that seem to still be having huge jumps in performance compared to desktop/laptop computers.
- binaryapparatus 9y agoSeriously, isn't this discussion of the wrong topic altogether? CPU is not too strong, battery is too weak by design. Am I missing the point in this and other articles arguing about everything else but not about battery capacity? What if we had 10% more battery capacity and (oh horror) 0.1 mm thicker phone?
- madeofpalk 9y agoYou mean what they did in the recent round of iPhones?
- binaryapparatus 9y agoMaybe. I hope so but we can't really tell if new batteries have security margin or not. It will show in 12+ months.
- WillPostForFood 9y agoI recently swapped the battery on my iPhone 6. I can tell you that if the battery is too weak it is due to space constraints. Mabe that's what you mean by design, but given the form and space available, I don't think the battery could be stronger than it is.
- binaryapparatus 9y agoSorry I wasn't completely clear. When I say 'by design' I don't really mean 'planned obsolescence', I mean 'system design problem' because somebody wanted to shave another 0.01 mm of thickness. I would be surprised to see any other reason than stupid form over function that causes this.
- DrScump 9y agoIf they knew this from day one, or even day 501, then insisting on designs with non-user-replaceable batteries looks even more like planned obsolescence for financial gain.