Post

I'm curious if your backend is targeting Wasm GC or linear memory.

GraalVM targets Wasm GC. We started with linear memory but GC performance was simply not good enough.

Do you know what the linear memory GC overhead was? Either compared to GraalVM's normal GC, or the browser's? (I would expect linear memory GC to have more overhead in some ways - e.g. stack scanning - but also the chance to use VM/language-specific low-level tricks)