The feature of numpy parallel can improve the efficiency of variable transform

[cfkstat]

When I transform large amounts of data, such as more than 2 million, the transform efficiency is relatively low, and the python package dpnp supports parallelism。

[guillermo-navas-palencia]

Hi @cfkstat.

Thanks for proposing this enhancement. I understand that transform for large datasets might be slow, but dpnp is an additional dependency which might only benefit Intel processors. Therefore, I discard its usage.