Results from dpbench of _pairwise_ workload (running on a laptop with 11th Gen processor and Iris Xe graphics): | workload | numpy | dpnp CPU | dpnp GPU | size | | :--- | :---: | :---: | :---: | :---: | | pairwise | 1.1 s | 4.4 s | 1.6 s | 16384 | demonstrates dpnp execution time on CPU which is 4 times greater than one of NumPy.