Expound upon "handle a lot of data" -- the applicable PDL module(s) depend on what operations you're performing on the data. See the linked PDL pod for a listing of the available PDL modules.
With regard to Tk::Zinc, if you reimplement the backend using PDL modules, you will see an appreciable speed (velocity is a vector) increase, presuming the size of the dataset warranted using PDL in the first place. The frontend won't be any faster, but the backend will make the data available sooner providing for more frequent updates.