Investigate: does copy_data function slow down the GPU version? #168
Description
in #126 we changed from a manual parallelised loop for copying data to copy_data from the standard library. Investigate if it is faster to do a parallelised loop for copying data especially for the GPU version.