UserAB1236872

I'll close then (Also, I don't believe C++ can be wrong. It's the reference implementation if I'm not mistaken?)

+1 This also should have a function to request the buffer length for purposes of allocating the buffer, however.

I'm taking a currently taking a (grad-level) deep network course which covers some practical implementation and optimization details. At some point I may be able to do some deep net...