When you say transpose, do you really mean transpose, as in to transpose a matrix, or do you mean transfer, as in let me transfer (copy) the matrix to the coprocessor? If you meant transfer, you might want to check out the use of array sections in offload transfers. The article https://software.intel.com/en-us/articles/xeon-phi-coprocessor-data-transfer-non-contiguous-array-elements-using-language-extensions shows how to do this, even if you have a non-unit stride.
