i just saw this thread on google and notice that they maybe someone like me who are searching for this so i post this
goodluck shonumi i hope that the emulator is as beautiful as dolphin
That's right, you can do anything you want in SW, but it's slow because everything is done by the CPU, that's why most emulators use it as an accurate "backend" which only handles 1x IR.
I never got my hands on this but I wonder if we can get better performance by using CUDA or OpenCL technologies.