I want to load my data into pandas dataframe and give it to gpu for processing. For this situation, the second approach is only using pydaal's tables and then giving to the GPU for processing.
If my data is homogenous, is there any benefit of using pydaal's tables for storage instead of pandas data frame ?
I am not going to use other functionality of pydaal. So basically my question is that merely storing your data in pydaal's tables has benefit on its own or not ?
The short answer to your question is no. Homogeneous Numeric Table for that purpose is equivalent to storing data in a C array, which you can also store within NumPy array.