Alright, well I didn't end up getting renderdoc working as I wanted, but I did solve my bug that I wanted to use renderdoc on in the first place. I ended up being able to do some profiling on my game engine after I got it resolved, and I'm reasonably satisfied with the results. My game engine isn't as fast as it'd be if I were doing stuff mutably, but it's got a reasonably good api and seems about as fast as the sorts of engines you can manage in a webpage now, which is about all I wanted, seeing as my usecase for it is solo-dev stuff and pumping numbers of active agents usually isn't what makes solo-dev interesting.

I did a rather simple program, a lot like bunnymark for pixi.js, and the performance seems comparable, with about ~6500 simple entities being able to be supported at 100 fps simulation, and the renderer can keep up at vsync, even when it's naiive (e.g. no batching). I use reducers for my game logic, so that ends up spreading the work out for simulation across most of the cores too.

that's cool as hell

Yeah it's been a lot of fun to design, and I've gotten inspired to pick it back up and work on it more with a friend in order to work on building a game on top of it. I think that'll be a good test of the API. Probably once I'm done with that I'll think about making some public content about it, like a blog post or something (I should really get my blog going).