Well, I think it's not good to split +0 and -0.
If you look at a processor with ~30 competitors.
If 15 participants bench +0 and 15 participants -0, there will be not much Points for this processor anymore.
And for example at most Socket A processors there are not much -0 participants. So there will be not much points for benching them subzero -> no motivation.
Another problem would be to proof that the result was made +0.
At some Processors you can see easily (Core i7 6GHz etc.)
but if you look at Socket A processors for example, some processors can reach frequencies under air/water easily that other chips need dice for.
Example:
http://hwbot.org/community/submission/1035565_benchbruno_cpu_z_athlon_xp_2500_barton_2785.4_mhz
(Dice, not good chip)
http://hwbot.org/community/submission/1042927_benchbruno_cpu_z_athlon_xp_2500_barton_2817.65_mhz
(water cooling, good chip)
there are even more extreme examples; I've seen ~3,3GHz Barton submissions with air cooling.
I hope I understood the new things right and I hope you understand my english