A great boost to see the full potential of Apple Intelligence.

  • @BorgDrone
    link
    English
    72 months ago

    And yet, I’ve never run into RAM problems on iPhones, both as a user and as a developer. On iOS an app can use almost all the RAM if needed, as long as your app is running in the foreground. Android by contrast is much stingier with RAM, especially with Java/Kotlin apps. There are some hard limits on how much RAM you can actually use and it’s a small fractIon of the total amount. The actual limit is set by the manufacturer and differs per device, Android itself only guarantees a minimum of 16MB per app.

    The reason is probably because Android is much more lenient with letting stuff run in the background so it needs to limit the per-app memory usage.

    Those apps also use more RAM than an equivalent iOS app, simply because they run on a garbage-collected runtime. With a GC there is a trade-off between performance and memory usage. A GC always wastes memory, as memory isn’t freed immediately once no longer in use. It’s only freed when the GC runs. If you run it very often you waste little RAM at the cost of performance (all the CPU cycles used by the GC) if you run it at large intervals you waste a lot of RAM (because you let a lot of ‘garbage’ accumulate before cleaning it up). In general, to achieve similar performance to non-GC’d code you need to tune it so it uses about 4 times as much RAM. The actual overhead depends on how Google tuned the GC in ART combined with the behavior of specific apps.

    Note that this only applies to apps running in ART, many system components like the web browser are written in C++ and don’t suffer from this inefficiency. But it does mean Android both uses more RAM than iOS while at the same time giving apps less RAM to actually use.

    It basically comes down to different architectural choices made by Google and Apple.