I need to copy a part of 2D Array to part of another2D Array i.e. implementation of
SLACPY from Lapack. In cublas library there are functions for setMatrix and GetMatrix,
but they are copying between GPU and CPU memory.
There is a low level cudaMemcpy2DArrayToArray routine with cudaMemcpyKind=cudaMemcpyDeviceToDevice
but I don’t know how to get the pointer to struct cudaArray from cublas.