There is no conflict, so you can do whichever you prefer.
For me, it varies. If there is a correlation matrix in a data set named CORR, I might read it into a matrix of the same name.
If the data set is named MyCorrelationMatrix, I'll probably just read it into S.
Most real data sets have both character and numeric variables, so you can't read all variables into a single matrix. For example: the sashelp.class data set contains a SEX and a NAME variable (both character), so if I want to read the numeric variables I might just say: