I did an estimation of this recently. It looks like even with 3 or 4 XHP35s in parallel, there is still probably too much current to each emitter. This is based on a rough extrapolation of the forward voltage from cree’s graph on their datapage and considering the voltage drop from the cell IRs and other circuit resistances. I estimate that an additional ~0.2 Ohms of resistance (like adding a coil of wire somewhere) would bring the current to around 2.5A to each emitter.