I wonder how they will tackle the infamous non-conformant Ps2 floating-point behavior issue, that is the biggest hurdle on emulating Ps2.

As of now, it looks like they're ignoring it:

https://github.com/ran-j/PS2Recomp/blob/91678d19778891b4df85...

   #define FPU_ADD_S(a, b) ((float)(a) + (float)(b))

But if you wanted to handle it, you'd presumably macro expand the floating point operations to something that matches the PS2 fpu (or comes closer).

Some context for others who were unaware: https://github.com/PSI-Rockin/DobieStation/issues/51

EDIT here's potentially a better link: https://www.gregorygaines.com/blog/emulating-ps2-floating-po...

PS2 floating-point behavior is one of the few hardware misfeatures so awful it affects emulation of competing systems[0]. The game True Crime: New York City is so dependent on PS2 floating point that the GameCube port installs an error handler just to make 1/0 = 0. Which isn't even PS2 hardware behavior. But it is "close enough" that the game does not immediately throw you into the void every time you step on a physics object.

[0] https://dolphin-emu.org/blog/2021/11/13/dolphin-progress-rep...

Probably the same way as the emulator themselves, with a list of titles needing the real PS2 floating point.

A lot of titles don't actually need it and work fine with standard IEEE floating point.