Honestly, I'm not sure if I should be impressed with the demo they showed. I haven't given the code a read, so take this with a grain of salt, but I feel like the exact demo they showed off was mostly a graphics card demo. You can get most, if not all, of that scene rendered purely on the graphics card itself using fancy shaders without needing to hit the CPU very much. I feel like I would have been much more impressed if they had created something such as a level from Megaman (since they showed a sprite of Megaman near the start) that the presenter was able to play. I would consider it more impressive since it would have to spend more time in Racket dealing with physics, collision, AI, etc. instead of spending the majority of the time on the graphics card. Hopefully they'll be able to show off a demo like that soon (if one doesn't exist already) to really show off the capabilities of the engine inside Racket. Honestly though, the whole project is exciting for me since they're really just trying to explore new ways of writing games. I'm looking forward to what they do in the future.

The presentation just before the "3D" one demonstrated a purely functional SNES-inspired engine, with sprites of Megaman, etc.

Relevant HN thread: https://news.ycombinator.com/item?id=8381274

It seems like the interesting thing is that it's a purely functional API. I don't thing they were trying to demonstrate that the language is fast.

Exactly. The engine is able to make excellent use of the GPU while sporting a purely functional API in Racket, a dialect of Lisp.

Right. I don't remember whether I said it in the talk, but that demo saturates the AGP bus, meaning that the engine can extract and write scene data at about 1.5 to 2 GB/Sec.

It's possible to put a lot more static content in each frame by uploading the vertex data to the card just once. That feature is on my to-do list.

As other said, it was a matter at bridging fast hardware with a lisp (racket) and its composable objects.