so is there a chance to get it fixed?
does skminer share the same code for each compute capability or are there different instructions that gets added depending on sm_?
if the latter it's just a matter of finding out where the bug is and fix it else it could be a cuda sdk bug that affects sm_30 and we're screwed...
i'm available for testing and debugging in case of you (or any other dev) not owning any sm_30 card.
thank you.

ps: could somebody tell if sm_30 cards are finding any block under windows?