Would prediction result change if the data was splitted into multiple smaller sets?

Hello, 
Thanks for the great package and I have used it analyzing some single cell datasets. 
But when I ran copykat on large-scale data (e.g., Xenium data with ~800,000 cells) it would not run sucessfully. 
I intended to split the data into some smaller sets, but I'm not sure it would give the same result as the whole dataset.
Do you have any suggestions?