Repository navigation
Replies: 1 comment 1 reply
|
@TFQ0 Thanks for digging into this. The ranking itself runs off the main thread, caches what it reads from titles and keeps its data size capped, so on its own it shouldn’t make the app feel slow. A fresh Home does need a few YouTube requests, but it shows your last feed first while those load. Looking into it I did find two startup things I’m fixing. The splash screen waits for the whole engine to load (and the first launch after the latest update also upgrades your saved data, so that one is slower). Also, the IzzyOnDroid and nightly builds were missing the app’s startup profile, which makes startup and scrolling slower on those builds. Could you tell me your device, where you installed Flow from (GitHub, IzzyOnDroid or nightly), and what feels slow exactly: opening the app, Home loading, scrolling, or starting a video? That helps me narrow it down. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Hi, and thank you for building Flow. It's an excellent YouTube alternative, and some of its features are even better than what the official app offers. However, I've noticed slowness during everyday use and would like to understand what might contribute to it.
Looking through the code, I noticed that FlowNeuroEngine scores and ranks candidates locally, while HomeFeedSources gathers related-video candidates through network requests. I also see that ranking runs on Dispatchers.Default, and that caching and prefetching are already implemented.
Could this combination of on-device recommendation processing and candidate-fetching requests contribute to the slowdown I'm experiencing, compared with YouTube's server-side recommendation infrastructure?
More specifically:
I appreciate the privacy benefits of keeping recommendations on-device. I'm mainly asking whether this architecture introduces a measurable performance trade-off, or whether the slowdown is more likely to come from another part of the app.
@A-EDev @my-live-is-a-lie
All reactions