
Retroarch's LPSR2 core (also known as PCSX2 core before rename) seems to have very promising results regarding latency!
So far i tested Radirgy/Ibara/Mushihimesama/Psyvariar 2 (Ibara and Mushi were tested because of their arrange modes, though having high quality music may make some people try playing on these ports instead of fbneo), and all of them have 2 to 3 frames of internal lag by default, alongside external latency being minimal, which is fairly weird as external latency made flycast unplayable and makes CV1000 games require runahead to work properly.