Data structure and performace

Hi,

I'm working on a report, which uses kpis from up to 10 different data sources. There are some data preparation steps I do in R. 

Most of the kpis share the dimensions, so my question is, whether its better to build 2-3 tables which contain all the kpis with shared dimensions or if each source should be represented by a table in power bi.

The highest priority for me is performance. Is there a huge difference between the two approaches? For the first approach i would use relations between the tables to be able to use slicers for the common dimensions. 

tldr: is there a big perfomance difference between few but wide tables and many but more narrow tables with relationships?

thanks in advance!