Benchmarking gradients + possible memory leak(?)

I’ve been interested in seeing whether react-native-skia supersedes react-native-linear-gradient and created a simple app to verify the performance characteristics of both.

My experimentation seems to have uncovered that react-native-skia keeps increasing the app’s memory usage and doesn’t seem to release it.

The conclusion indicates that react-native-linear-gradient is much more performant for this use case, with skia also having a weird artifact where views appear one at a time.

This is react-native-linear-gradient with 1000 views rendered. https://user-images.githubusercontent.com/697707/175272980-5406e356-198e-45a7-a840-ab7b9c72c44c.mp4

This is react-native-skia with 1000 views rendered. (notice how memory increases) https://user-images.githubusercontent.com/697707/175274506-234d670f-c88c-4f1c-a360-691df52514d2.mp4

The code is available here: https://github.com/andreialecu/rn-perf-gradients

[wcandillon]

We believe to have fixed the issue, I will send you a test package and maybe you can confirm it?

[wcandillon]

@andreialecu Thank you for bringing this to our attention in such a compelling example. I am closing this issue was it’s mentioning two things that we are tracking separately:

• #628 for the memory leak
• #610 for the slow startup time

But please feel free to reopen if needed.