1. Anuncie Aqui ! Entre em contato fdantas@4each.com.br

[Flutter] Impeller Rendering Engine Causing Massive Latency and Memory Consumption

Discussão em 'Mobile' iniciado por Stack, Outubro 10, 2024 às 11:22.

  1. Stack

    Stack Membro Participativo

    I'm working on a Flutter project (sdk: '>=3.0.1 <4.0.0') for iOS (platform :ios, '13.0') and right after upgrading Flutter this morning I noticed that the iPhone simulator (Xcode 14.3) was using between 3-5gb of memory and my entire system was unusable as long as the simulator was running.

    I have a 32GB Mac and when the simulator was onscreen (when minimized the system was performant and seemed fine) and running, my entire system was experiencing massive latency, lags, and performance issues. Scrolling through source, doing any kind of debugging, or really anything at all was extremely difficult due to the performance impact (VS Code for my IDE).

    After spending several hours debugging what the heck was going on all of a sudden (this had never happened before the upgrade), I noticed in my debug console a line I hadn't seen before...

    "[VERBOSE-2:FlutterDarwinContextMetalImpeller.mm(35)] Using the Impeller rendering backend."

    After some quick Googling and reading about how Impeller is apparently so awesome and incredible, I decided to turn it off and see what would happen. I added the following tag to my info.plist to disable it.

    <key>FLTEnableImpeller</key>
    <false />


    Low and behold to my delight, but certainly not surprise, everything went back to "normal" and I was able to actually scroll through a source file while the simulator was running my project on screen at the same time! Almost like I wasn't in the early 1990's anymore working on my 386... Simulator memory was at around 140-500mb at most.

    My question is, how can I enable Impeller and still be able to use the rest of my system at the same time? Has anyone else experienced this behavior or had a similar issues with Flutter/Impeller after upgrading or at any point? Would love to hear peoples thoughts on this.

    Continue reading...

Compartilhe esta Página