I’m having some C++ issues.
Everything works, but it isn’t looking too good for performance.
In particular:
- I have a 3-element vector class that uses three cuDoubleComplex elements (which I operate on with casting to another class, zcmplx [cannot use straight zcmplx due to another compile-time error]). I can use my zcmplx class fine, but the 3-element vector one incurs quite a bit of local storage use.
- The compiler complains a lot about *this references (it’s not sure where in memory they point to).
Are there any PDFs that document the Cuda C++ implementation?
Ben