I'm not sure about a 4090 but there is a fork for 3090s: https://github.com/Don-Chad/ninfer-3090

I'm running it using a 4090 on using llama.cpp with Q5_K_S and its running at ~33 t/s