It's probably too late, but in first place read more precisely into the description, because currently your solution is wrong.

C++ version of kata should use more modern and familiar type for passing board to function. Passing fixed arrays is very unusual to C++, and if we taking into account that it is actually a pointer to fixed array it becomes even more non-obvious, unsafe and inconvenient to work with. I think there is should be some container from STL, such as std::vector< std::vector< unsigned int > > or std::array< std::array< unsigned int, 3 >, 3 >.

I think this kata is completely broken for C++.

I tried two very diffirent approaches, and they both don't pass tests and return false in some case(s), where expected true (but they both pass random tests though and any test on my local machine).
And yes, I know about elements ordering, possible values duplications(which was not obvious from instructions) and negative elements appearing(which was absolutely surprising for me and broke up one of my solutions at first)

Maybe I'm missing some very important detail, that not are obvious from instructions.
I would appreciate if you point me on this detail, but right now I think this kata is broken in some strange way. And I can describe solutions in further details or post code of them, if someone needed