Inplace

CUDA and OpenMP implementations of the C2R and R2C inplace transposition algorithms. These algorithms are described in our PPoPP paper.

We have included a specialization for very tall, skinny matrices that yields good performance for in-place conversions between Arrays of Structures and Structures of Arrays.

The code includes OpenMP and CUDA implementations. The OpenMP implementation is declared in <inplace/openmp.h>, while the CUDA implementation is declared in <inplace/transpose.h>, and carries the following signatures:

namespace inplace {

void transpose(bool row_major, float* data, int m, int n);
void transpose(bool row_major, double* data, int m, int n);

}