Multiple imputation in group R: mice vs mi

Question

What are the advantages and disadvantages of these two packages for multiple imputation?

Here is my motivation for asking this question: I have used the mi package so far, but I cannot get convergence. My dataset is about 10,231 x 28. Four variables have 18-20% of the missing values, one has 14% and the rest is about 5% (or lower). There is no convergence even after working on the server for 18 hours. Now I am wondering if mice work better.