But why does that make a difference? Regardless of whether you reverse engineer an API or have the full documentation available, when you translate from it to something else, you're building the same thing: an API translator. It doesn't change whether it's an emulator.

Traditionally, "emulator" refers to hardware emulation. APIs are reverse engineered, mocked, translated, etc. all the time, but we don't call it "emulation".

Even simply mimicking the behavior of a unified virtual memory space is not called "emulation", unless there are hardware-specific quirks that need to be emulated (which may very well be the case for PS5 caches, I don't know).

I'm trying to understand, what is being emulated?

You said it's emulation because it handles ps5's graphics APIs and unified memory. I noted that proton is also translating vulkan to another graphics API. You said it's different because the api we're targeting is proprietary and needs to be reverse engineered, implying that is the distinction. Right?

So then doesn't that imply that if we did the same thing, but without needing to reverse engineer anything, it would now be an apples-to-apples comparison with what proton is doing?